Apple Valley, California, in the heart of the desert.
You walk out here, you don't see street lights, you see stars. Here, it is different. Colder in the winter, hotter in summer.
It's an every day trip for glass fitter Robert Smith.
You can put glass into anything that moves - from shower doors to store fronts, cars. I've yet to do an aeroplane, but I have done a train.
But his journey over the past 30 years has been something of a miracle. At the age of two he was diagnosed with severe autism.
Hi, Bobby, how are you?
This is Robert, aged three. He refuses to make eye contact, a key sign of autism. He appears cut off from the outside world. He is preoccupied with repetitive play, characteristic of autistic children and he has no speech. This is the first time mother and son have seen these images.
It's pretty crazy. But we made it.
Back in those days, what problems did Robert have?
He would rock himself and hit his head against the wall, repeatedly. I began to hear some children with autism would wear helmets, to prevent brain damage or whatever. So I kept him out of the crib as much as I could. He had other behaviours. Moving his head back and forth with his hand in front of his face. Making a croaking noise.
Yeah, he would sit and do it, do it.
I don't remember half of it. To me it seemed like a place where I was going to do things. It was almost like school, you know. It doesn't register much, I don't remember feeling any different. It was just a stage in life.
Do that. Yeah! That's what I'm looking for! Do this. Yeah!
This is the therapy technique that cured Robert. It's now practised across California.
Leigh is teaching Luke to imitate and perfect imitation skills. She gives a cue, Luke responds and she provides positive reinforcement.
Do this. Yeah! Thank you.
Tasks are broken down into simple components. Success is rewarded with praise, such as chips or a drink. It's known as Applied Behavioural Analysis or ABA.
We've had a lot of success stories. Success ranges from children just learning to talk to children now in third to seventh grade, independently without assistance. They're in regular ed without help.
The hills above Los Angeles is the home to the man who pioneered ABA for autistic children. For 40 years psychologist Dr Ivar Lovaas has developed the approach at the University of California.
SR IVAR LOVAAS:
Imitation is a major learning strategy. Half the kids learn to imitate vocal sounds, like "mama," "papa." Then "mommy, daddy", and then slowly you make it more complicated. It's like an accelerated curve. You take a long time to get to the first sound imitation, then they shoot up like this. So, you can say, "What's your name?" and they can say "Adam."
In 1987 Lovaas published the results of research on a group of 19, mostly from around Los Angeles, who'd received 40 hours intensive ABA.
At the age of seven, 47% of them scored within the normal range, or the typical range on IQ tests and were in first grade unassisted. We tested them again when they were 13, as adolescents. The best outcome group maintained again which surprised me. I thought it really meant in those first years that we had them in treatment they learn how to learn.
Marty, touch your nose. Good boy.
The ABA team at the university of California had to swim against the tide. Many of Lovaas' rivals believed autism was created by a cold mother-child relationship, and as such was treatable with Freudian psychoanalysis. That's now discredited but some researchers still argue against intensive ABA. This is the alternative, the classroom based approach. Ivy and Jared spend all day in class together with other autistic children.
Catch the spider.
Along time rival of Lovaas believes this is the best model. He says that 40-hours of intensive one-on-one therapy can actually be damaging.
DR EDWARD RITVO:
Children with autism need other things be sides one-on-one relationships. They need to play and to be kids. They need social reinforcement. I've seen kids who have been placed in this treatment 40, 50 hours a week, day after day, week after week, and they lose out on the part of their personality that could develop and become more social.
Now Lovaas says the technique is not effective unless have you something like 30 or 40 hours intensive ABA per week. What do you say to that?
With two or three hours a day, four or five times a week, we seem to reach the maximum effectiveness.
There are no data to suggest that two or three hours a day is effective. The easy thing is to say, "Oh no, your child needs... is to be stressed now, only need two or three hours a day and then play with other kids." They don't learn anything by being with other kids. We know that for a fact. You put a class in a normal class, if he doesn't learn the basic skills he will stay the same.
ABA is not for everyone. Standards of treatment can vary. Jared's parents tried it at home with a team of therapists. Their son made little progress.
The things you hoped to see such as increased eye contact, more self awareness, decreased aggression, we were not seeing. That he was non-responsive when it came to that treatment. Truly it was after we took the somewhat unusual step of wiring his room with video to watch the therapy that we were aghast at what we saw.
Clap your hands. I don't like. I don't like that. Clap your hands.
Jared cries as he fails and fails again Lovaas' own data shows that a minority of children don't respond, even to the best ABA and much depends on the quality of the therapy.
You need to listen. Stand up. When I say stand up, you stand up.
Dr Lovaas acknowledges that the majority of therapists fall short of his standards. His critics argue that ABA is overused and overhyped. Despite the controversy, there is a growing body of evidence in California that early intervention does work actually work if it's intensive and early enough. The crucial question is what the state believes is the best way forward. I have come to meet the man in charge of the $1.8 billion budget for autism. California's Department of Developmental Services enjoys a big budget for disabilities guaranteed under state law. The state's most senior psychologist says he is ready to embrace a radical new policy on ABA.
DR RON HUFF:
ABA definitely works. It probably has more science behind than any other approach. I think every child should be given the opportunity to find out whether or not that child can respond to ABA. Now, some children...
That's significant. That's a big financial resource?
It is. It is, but we are looking at a lifetime of developmental problems if the child does not receive.
The situation in California is very different from in Britain. Here state politicians seem willing to back ABA despite the huge costs. The costs will be vast. Here, there is an epidemic of autism. Numbers up are 300% over the last decade. The crucial question is why. According to an official California study. The number of causes diagnosed was consistent between 200 and 300 a year in the 1970s. The numbers climbed sharply and climbed during the 1980s. Today, more than 3,000 cases a year are being diagnosed. There are only two possible reasons for. This either some unknown environmental trigger is creating an increase or the rising numbers are a by product of better diagnosis.
Instead of autistic kids slipping through the cracks, they're getting all the gravy now. It is very advantageous to have their kid labelled autistic because they get a nice teacher and ABA and all goodies from the school system. Autism is well funded as a disease now, thank God.
You think that re-categorization could account for 100% of this massive rise?
Russell is one of 18,000 autistic children in California today.
Who is the best?
You had better believe it.
Swimming is one of the few skills Russell has learnt since he was diagnosed autistic aged two.
You experienced clearly a feeling of death and remorse in your family when you see someone that you love more than life itself disappear before your eyes.
Like many parents, Rick is convinced that Russell was damaged by a series of vaccinations. He strongly rejects the idea that the epidemic of autism can be entirely explained by poor diagnosis in the past because numbers have rose over the last few years.
Missing child with autism is like missing a train wreck. For us now to now think that somehow we have better identified a child who can't talk, who has repetitive behaviour. Who makes no eye contact. Who is self- involved and in many cases self-abusive just defies logic.
Is it credible that such a massive rise in numbers can be put down solely to changing diagnostic practice?
I don't think can you. I think we would be foolish attribute that rise simply to one single factor. If we are intelligent, I think we are doing this the intelligent way, we are looking at all of those issues. If it is determined that an environmental link is there, we are going to see a lot more of this before we can correct that. Because we have been contributing things to the environment for many, many years. I don't know if we can clean that up.
A ten year study into possible links between autism and environmental pollutants and vaccinations is now under way. Meanwhile the case load continues to swell. With ABA, the treatment of choice, costing $50,000 per child, per year, the budget for special needs is already stretched.
If the child was not treated then the child would be in an institutional setting and protected care for 40 years or 50 years. Normal life expectancy are 60 or 70 years. That costs the state $30,000 a year. You multiply that with 50 and you are coming into millions of dollars.
Back in the desert Robert Smith prepares for the night shift. He was one of nine children to completely overcome their autism in Dr Ivar Lovaas 1970s research study. Now he makes $50,000 a year as a glazier. The same as a year's worth of ABA. California is able to back ABA for every autistic child. The hope is many more will recover.
This transcript was produced from the teletext subtitles that are generated live for Newsnight. It has been checked against the programme as broadcast, however Newsnight can accept no responsibility for any factual inaccuracies. We will be happy to correct serious errors.