Most of us do it without even thinking about it.
Whether we're looking at someone we know, or a person walking down the street, or even a photograph in a magazine, we often immediately try to figure out what the person is thinking or planning.
Is she looking at her watch because she wants to know the time, or is she trying to break off the conversation? Did he look up because something moved overhead, or because he's amazed by something I said?
Psychologists call this "theory of mind" -- that is, figuring out what is in someone else's mind -- and they say we start developing this skill as babies, before we learn how to talk.
But there is one group that is particularly poor at theory-of-mind cognition -- people with autism. And now, studies at Carnegie Mellon University have shed new light on what may be malfunctioning in autistic people's brains.
One of the hallmarks of autism is difficulty in social relationships. Children and adults with autism often have trouble making eye contact, interpreting facial expressions and behaving appropriately in social settings.
These behaviors can often embarrass parents and anger strangers. In his book "Unstrange Minds: Remapping the World of Autism," George Washington University anthropologist Roy Richard Grinker described how his autistic daughter had walked up to a female stranger and snapped her bra strap, and had once asked a rotund man, "Is there a baby in there?"
The Carnegie Mellon studies suggest that one of the key problems that underlie such behaviors is that the brain areas that do theory-of-mind processing are badly connected in people with autism.
Researchers tested this most recently by asking a group of people with high-functioning autism and another group that did not have the disorder to watch three types of line-drawing animations while lying in a functional magnetic resonance scanner, which can measure blood flow in the brain second by second.
In one set of animations, the shapes move around randomly. In another, they dance in a coordinated way.
But in a third set, they interact to tell a little story. In one case, a large triangle tries to coax and then nudge a small triangle out of a circle, and then seems to congratulate the small triangle when it finally leaves the nest.
At least, that is the interpretation that most of the non-autistic people came up with.
The people with autism had a much harder time figuring out what the animated characters were doing and why, said Rajesh Kana, the lead author of the paper and now a psychology professor at the University of Alabama at Birmingham.
Brain images taken while the autistic people were struggling with this task showed activity in one key area of their theory-of-mind network, a spot above and behind the right ear called the temporal-parietal junction, but also showed that other important spots in the frontal lobe of the brain, behind the forehead, were less active than in the control group and were not synchronized well with areas farther back in the brain.
The Carnegie Mellon researchers believe the frontal areas in the network help analyze intentionality, and because they aren't coordinated with the other parts of the network, autistic people are left at a disadvantage.
"Imagine a quarterback throwing the ball, but the guy he's throwing it to keeps dropping the ball, and eventually the quarterback says, 'I'll just run the ball myself,' " Dr. Kana said.
"In a similar way, in people with autism, the back of the brain is throwing the ball to the front of the brain but it keeps dropping it, so eventually the back regions decide to do the job by themselves, and what happens is they might click on the right answer one in 10 times."
This "functional underconnectivity" between the front and back of the brain is a general problem that several studies at Carnegie Mellon's Center for Cognitive Brain Imaging have found in autistic people on everything from language comprehension to memory tasks to the theory-of-mind tests, said the center's director, Marcel Just.
"For instance," he said, "people with autism have difficulty with interpreting complex sentences. However, they have an advantage over other people in reading individual words, and often have superior spelling abilities."
Brain scans done while the autistic people were reading showed two parts of the language comprehension network were active, but a third area wasn't. "Over and over this pattern shows up" in the thinking problems that autistic people have, Dr. Just said.
One of the first to discover the difficulties autistic children have in understanding what others are thinking was Simon Baron-Cohen, a professor of psychopathology at the University of Cambridge.
He uncovered the problems using a standard exam known as the Sally-Ann test. In this story, Sally puts a ball in a basket, then leaves the room. Ann then takes the ball and puts it in an adjacent box. Children are then asked to say where Sally thinks the ball will be when she returns.
Dr. Baron-Cohen found that most children, including some with Down syndrome, understood that Sally would look in the basket for the ball, because that is where she left it. But the autistic children thought she would look in the box, because that is where the ball actually was.
Rebecca Saxe, a cognitive scientist at the Massachusetts Institute of Technology, has used adult versions of the Sally-Ann test to help determine what part of the brain was being used to figure out other people's beliefs.
She scanned people's brains while they carried out two tasks.
In the first, they were told that Susie had driven her sports car home and left it in the driveway. During the night, Nathan moved it into the garage to make room for his minivan. They were then asked to say where Susie would expect the sports car to be when she came out the next morning.
In the second task, they were told a volcanic island had experienced an eruption three months before that wiped out all the vegetation on it. They were then told that satellite photos of the island were taken before the eruption, and asked whether they would show a rocky landscape or one covered with plants.
The two tasks were the same except for one thing: in the first, people had to interpret what was in a person's mind. When Dr. Saxe subtracted the brain areas involved in the volcanic island task from the sports car task, she found one unique area involved in theory of mind -- the right temporal-parietal junction.
Dr. Saxe has done newer studies showing that this area also lights up when people are making moral judgments, particularly when they are deciding how honorable someone's intentions are.
At a Kavli Science Journalism workshop at MIT in June, Dr. Saxe said it makes sense from an evolutionary standpoint that a large part of our brains is devoted to our social relationships.
"In the human environment, the major cognitive challenge is other people. We live in large groups. Our survival depends on social relationships with groups, from birth to death."
Still, Carnegie Mellon's Dr. Just was surprised to learn how much of our mental energy goes into social cognition. When he started doing brain imaging, he said, "I focused on arithmetic, sentence comprehension, chess, because I thought that was real thinking. But it's amazing that there are these other networks that are very frequently active that are doing various kinds of social computations."
When he tested autistic people and non-autistic people on how well they remembered faces, for instance, he found the temporal-parietal junction lit up in the non-autistic group, but not the autistic group.
"There was no theory-of-mind challenge there -- it was just, tell me if this face was the same as the face you saw before. But I think this reflects the fact that when the people with autism were looking at the face they were thinking, 'Oh you know, wide cheekbones, broad nose,' coding it as a visual object, while the control subjects were adding to it [such aspects as] friendly, intimidating -- some kind of social interpretation."
And while he thinks the temporal-parietal junction is important, he believes it is the lack of coordination between that zone and frontal areas of the brain that may be the real problem for people with autism.
That finding isn't just a new way of understanding what might be going on in autistic people's brains, he said, but offers a possible avenue for therapy.
Just as his center's research has shown that children with reading problems can improve the reading network in their brains after intensive therapy, the same thing might be possible for many with autism.
"In the same way a reading teacher might repeatedly train a student in word recognition, maybe an autism therapist might repeatedly train someone in inferring an intention. It sounds too simple, but that's what it may come down to."
Correction/Clarification: (Published Sept. 19, 2008) Roy Richard Grinker is an anthropologist who wrote a book on autism. His first two names were transposed in this article on autism research as originally published Sept. 17, 2008.
Mark Roth can be reached at email@example.com or 412-263-1130. First Published September 17, 2008 4:00 AM