As concern over autism grows, one of the most hopeful signs for families is that early therapy seems to help many children overcome the worst of their verbal and behavioral deficits.
Right now, though, there is an obstacle to how soon that therapy can begin: Autism usually cannot be diagnosed firmly until a child is 2 or even 3 years old.
University of Pittsburgh researchers Mark Strauss and Jana Iverson are hoping to shorten that delay drastically with the studies they are doing on detecting the early signs of autism.
Many parents become concerned about their child's development around the age of 1 or even earlier. Their baby doesn't look up at them when they approach the crib. He has trouble sitting up. He is obsessed with playing with one toy.
But they don't know for sure if this is a sign of trouble, the researchers say.
"We're trying to tighten up that two-to-three-year gap between when a parent knows something is wrong and when a diagnosis is given," Dr. Iverson said, "so that this valuable time for intervention isn't wasted."
The most severe forms of autism are marked by language delays, difficulty making social contact, and repetitive behaviors, sometimes accompanied by retardation and seizures.
At the other end of the spectrum, higher-functioning autistic people often have an above-average intelligence and vocabulary, but are socially awkward and obsess over specialized interests.
The two psychologists are taking different approaches to the early detection problem.
Using an infrared camera that tracks eye movements, Dr. Strauss is studying how babies react to pictures and videos of people, patterns and objects.
Dr. Iverson's team visits babies at home to watch for such key developmental signposts as babbling, crawling and walking, and how they interact with their parents and others.
Both are working with the same type of children -- the younger siblings of children who already have been diagnosed with autism.
While the current national estimates are that one in 150 children is born with some form of autism, the autism rate among younger siblings of autistic children may be as high as two out of 10.
The Pitt psychologists are part of an international team of researchers sponsored by Autism Speaks, the major advocacy organization for autism, known as the High Risk Baby Siblings Research Consortium, which is hoping to pin down the earliest behavioral and biomedical markers for the disorder.
Dr. Strauss' research is based on his earlier work, which showed that children and adults with autism struggle with remembering faces, reading people's moods, deciding which gender someone is, and even putting animals and objects in the right categories.
Most of us start learning how to do these things as babies, without ever being conscious of it, he said.
"We learn a tremendous amount of information about things we call implicit knowledge -- so you learn how cats and dogs are different, or you figure out how to recognize faces -- and people think about this as being trivial because it's so simple for us."
But the mental processes involved in these tasks are actually very complicated, Dr. Strauss said.
"They have computers across the street that can play chess at the master's level," he said from his office in Oakland. "They don't have computers that are good at face recognition, discriminating gender, or recognizing categories of objects. These are things we do automatically, but they're actually quite complex."
One experiment he has done is to show people pictures of men's and women's faces, some of which are distinctively masculine or feminine, and some of which are more ambiguous.
The people in the photos do not wear makeup and put on backward baseball caps to obscure their hairstyles, yet most adult observers have little trouble picking up the sometimes subtle differences between the male and female faces.
Adults with autism struggle with that challenge, however. While they get better at it, people with autism still are only about 80 percent accurate in identifying whether a "crossover" face is male or female, he said.
The same thing happens with categorizing animals and even objects. People with autism take much longer to identify whether an odd-looking cat or dog belongs to its animal group, and while they have no trouble putting a straight-backed, four-legged chair in the right category, they struggle with how to classify a love seat.
While no one knows for sure why this happens, Dr. Strauss' theory is that most of us create typical composites of faces, creatures and objects that we use as reference points when we look at the world, while people with autism cannot do that easily and tend to concentrate on parts of a face or object rather than the whole.
One way to appreciate the effort people with autism have to make, he said, is to imagine what would happen if a non-Asian had to remember the faces of people he had met the night before at a cocktail party in China.
Studies have shown many Americans have trouble telling Asian faces apart because they didn't grow up in that culture. So in the cocktail party example, he said, in order to remember a face, "you're going to try to start memorizing details -- this person had a mole, or that one had glasses -- whereas in your own culture, you wouldn't have to do that.
"Well, the person with autism is always doing what you're doing in China."
The challenge in Dr. Strauss' current research is how to do experiments with babies who haven't learned to talk.
The answer is a camera that can track the infants' eye movements when they look at videos or photographs. If the baby is looking at a face, the system's software shows the location of the baby's gaze as a red dot that gets bigger the longer he stares at that spot.
Child development experts already know that babies will look longer at a face or object that is new, or that is especially attractive, and Dr. Strauss hopes to use that tendency to see if there are distinctive differences between normally developing babies and those later diagnosed with autism.
While Dr. Strauss' work is focused on his Pitt laboratory, Dr. Iverson's group observes babies in their homes to look for signs of delayed development.
She is particularly interested in three things -- motor skills, babbling and talking, and what psychologists call "joint attention" -- when babies pay attention to their parents or try to get the parents to pay attention to them.
In a previous study of 20 younger siblings of children with autism, she was able to detect sharp differences in babies who developed the disorder and those who didn't.
A video clip of an unaffected baby showed her crawling around the room, babbling and holding up a toy for her mother to see. A clip of a child diagnosed as autistic showed him sitting still, focusing intently on one toy and not paying any attention to his mother's words.
The good news from that pilot study is that 18 of the 20 siblings did not develop autism, and she expects similar numbers in her current project. "Most of them will be fine, absolutely. There may be some quirky things about them" on their social interaction, she said, "but who's not quirky?"
And even though her paired video clips show striking differences in the two babies, she said, "I don't think at the end of the day we're going to find that there is any one trajectory that leads to an autism diagnosis."
The kinds of delays she has seen so far include delayed babbling and speech and a reduced vocabulary; trouble sitting up and later, crawling and walking; and shortcomings in social communication -- "not holding up a toy for someone to comment on, not pointing, not really engaging another person around an object of interest."
The affected babies "were extremely content and happy children," she said, "but just not very interactive."
One reason to be cautious about all these symptoms, though, is that many babies show these kinds of delays early on, but are normal by the age of 2, whereas a smaller group develop normally for the first year of life, and then regress somewhere between 18 months and 3 years old.
The babies that regress into autism later on can be especially heartbreaking cases, and they have fueled much of the speculation that childhood vaccines cause autism, a theory mainstream scientific researchers have rejected.
But in Dr. Iverson's view, it's the children who overcome early delays who are more fascinating.
"There are these kids who have early delays and we've watched them and then at 2 years old they're terrific, and you think, where did that rebound come from?" she said.
Dr. Strauss would like to study 150 high-risk children over the next five years, but so far has only been able to enroll 10. Dr. Iverson wants to assemble 160 children for study, and to date has enrolled only 20.
One problem both face is that many parents who have a child with autism decide not to have any more children.
"We're looking for needles within needles in haystacks," Dr. Iverson said.
Wednesday in Health: Brain scanning studies at Carnegie Mellon University show wiring problems in autistic people's brains.
Mark Roth can be reached at firstname.lastname@example.org or 412-263-1130