(CNN) Children with autism spectrum disorder often have either an exaggerated or a numbed response to sight, sound and touch. This behavior is so common that it's one of the diagnostic criteria for the disorder.
Now, a new study suggests that children with autism might also experience smells differently from children who have typical development.
To explore how children with autism respond to odors, researchers built a device called an olfactometer, which delivered different scents through a small tube that fit into nostrils. A second tube measured how much air the children were breathing in during each scent -- in other words, how much of a sniff they were taking.
The researchers exposed 36 children -- 18 who had autism and 18 who did not -- to alternating bouquets of pleasant smells, such as roses or shampoo, or unpleasant smells, such as sour milk or rotten fish.
They found that children who did not have autism took a longer sniff for roses and a shorter one for rotten fish, and the difference in breathing happened quickly, within one-third of second of being exposed to the scent.
In contrast, children with autism did not change their breathing in response to the different aromas.
"What we measure, the sniff response, is quite intuitive (and) adults and children with typical development react similarly," said Liron Rozenkrantz, a doctoral student in neurobiology at the Weizmann Institute of Science in Israel. The difference found in children with autism was striking, said Rozenkrantz, who was lead author of the study published Thursday in Current Biology.
It is unclear from the study if children with autism did not have a sniff response because they didn't perceive odors in the same way other children did, or because they just couldn't control their sniffing behavior.
Still, the idea that differences in smelling could be a symptom of autism might not be news to parents of children with autism. There have been accounts from parents about children who want to sniff people or objects, or who choose their friends based on their odor, even when it's imperceptible to others, Rozenkrantz said.
It has been difficult to study smelling behavior in children with autism because tests often ask them to describe odors, even though difficulty in communicating is a hallmark of autism, Rozenkrantz said.
The olfactometer-based test takes about 10 minutes and does not require the children to answer any questions or complete a task. In the study, they wore the device while watching cartoons.
The researchers in the current study found that their olfactometer could accurately identify the autism status of 12 of the 18 children who had autism, and 17 of the 18 control children who did not have autism.
The ability of the device to correctly categorize autism in 81% of cases raises the possibility that it could one day be used as a way to diagnose the disorder. The fact that it does not require a verbal test means that the olfactometer could be used in babies as young as a few months old, Rozenkrantz said.
However the average age of children in the current study was 7. For the device to become a method for early diagnosis, it would first have to be tested in young babies, and babies identified as being at risk of autism would have to be followed into childhood to see if they actually developed the disorder, Rozenkrantz said.
Research has shown that early behavioral interventions for young children at risk of autism could improve their development. Most children are not diagnosed until age 4 or older and have missed a key window to get help.
"It's way too early to say whether this could be helpful in diagnosing autism," said Dr. Paul Wang, senior vice president and head of medical research for Autism Speaks, an autism research and advocacy organization. Wang was not involved in the study.
In order to be used as a diagnostic tool, the device would have be tested in more children and also show that it can identify autism in more than two-thirds of children with the disorder, he said.
However the current study does give some idea of olfactory symptoms that could be occurring with autism, Wang said.
"I think this really fits well with increasing research on general sensory symptoms in autism," he said.
The researchers saw that children who had a dampened sniff response were more likely to also have deficits in communicating and socializing, but were less likely to have repetitive behaviors, which are the other hallmark of autism.
Differences in sensory reactions have previously been thought of as part of the repetitive behaviors of the disorder, but "this is an interesting study that suggests that the senses are more tied to socio-communicative behavior," Wang said.
These findings could also provide some clues as to why many children with autism have trouble eating.
"It could be muscular problems, with swallowing and chewing, but smell is a big part of taste and children with autism may not perceive smell in the same way as other children," Wang said.