What the study's authors (Timothy P. L. Roberts, Ph.D., J. Christopher Edgar, Ph.D., Deborah M. Zarnow, M.D., and Susan E. Levy, M.D.) did was to have 30 autistic children ages 6-15, and 34 age-matched NT children, listen to recordings of fairly simple, neutral sounds --- tones of varying frequencies (probably like what you hear when you get your hearing tested) or a voice, either speaking complete sentences or pronouncing random English vowel sounds. The children were not told to do anything, or to listen for anything in particular; just to listen, while sitting under a biomagnetometer***, which measured the magnetic fields generated by their brains as they listened.
While both groups of children appeared to be using their superior temporal gyri to make sense of what they heard, the autistic children distinguished themselves by responding more slowly and to a lesser degree than their nonautistic peers.
In particular, they were looking at the M100 latency, which is the lag between the 100-millisecond mark and the characteristic peak in the wave-form expressing the strength of the magnetic field generated by a particular area of the brain (in this case, a spot on top of the temporal lobe, toward the middle of the brain). The length of time it takes for this peak to form has been shown to vary in relation to certain properties of the sound being heard, as well as between the hemispheres of the brain. In this study, the range of M100 latencies for autistic children was about half that of NT children; the NT children's M100 peaks might occur anywhere within a 40-millisecond window depending on the frequency of the tone they heard, while the autistic children's responses were closer together, staying within a 20-millisecond window. Also, when two or more tones were played in quick succession, the autistic children's response to the second tone would be diminished, and when a long series of similar sounds was interrupted by one sharply contrasting "oddball" sound, the autistic children lagged their NT peers by 50 milliseconds in detecting it. (I am assuming they equate detection of a sound with altering one's brain activity in response to it). The abstract also mentions "abnormal patterns of gamma oscillation" in the superior temporal gyrus of autistic children, but without a full article to refer to, I cannot guess what they might mean by that.
Reuters quotes Dr. Roberts on the potential diagnostic utility of these findings:
Obviously, those same patterns would have to be observed reliably in 1- and 2-year-olds before MEG could be used as an early diagnostic instrument; as of now, I think MEG studies have only been done in older children.
Children are usually diagnosed with autism only after they reach age 2 years or older and Roberts said the hope is that MEG could diagnose children as young as 1 year, so therapy could begin earlier and perhaps be monitored to evaluate the results on the brain.
MEG can cost roughly $400 an hour to perform, but it is harmless and could become less expensive if more devices were available. MEG is used currently to help locate brain tumors and to diagnose epilepsy.
Roberts foresees MEG being employed to examine people with attention deficit disorder and other mental problems.
He said it may also provide researchers with more clues to the causes of autism and help solve the dilemma of what is hereditary and what is environmental about the condition.
Kristina Chew wonders how, if this ever did come to be a widespread diagnostic tool, doctors would tell the difference between autistic toddlers and hearing-impaired ones. While I think deaf children, or children who couldn't hear as wide a range as most people can, could be easily screened out (they'd fail to respond at all to some or all of the tones, as opposed to responding later), children with auditory processing disorders (that make them less able to distinguish between sounds), tone deafness or some other more subtle hearing or cognitive disability might be mistakenly ID'd as autistic by this metric. I think the solution to that problem lies in a very careful reading of the results, in the use of multiple types of imaging, and the reliance on multiple indicators. (Another potential early-identification signal for autism, much ballyhooed a year or two ago, was gaze cuing --- tracking eye movements as the child watched a video of a human face to see if they met the actor's eyes, which NT children apparently learn to do around six months old). Just with regard to hearing, Autism Diva has an impressive roundup of what's been discovered about autistic differences; indeed, one of the things she cites is a huge difference between the left and right hemispheres in M100 latency. Another autistic difference she cites is superior ability at identifying individual notes within a chord; it is highly unlikely that a child who consistently shows both a delay in processing sounds and greater perceptual acuity is hearing impaired.
Basic information on MEG can be found here and here; a Powerpoint presentation on the finer points of using MEG to measure auditory-cortex activity in response to different types of sounds can be found here; for a paper on how different vowel sounds affect M100 latency, go here.
*The full article didn't make it into the print edition, which is what I actually read every day. No, there was only the tiniest stub of a notice there --- maybe two or three sentences in a box on the far left side of Page Two. The print edition of the Star, like those of many other newspapers, is shrinking rapidly, and opting for a much lower content-to-fluff ratio on the few pages it still prints. This is very annoying if you, like me, prefer to read rather than hear or watch your news** and find staring at screens irritating.
**Ironically, the very research I describe here deals with why I --- and, quite possibly, you too --- might have this preference.
***The biomagnetometer is described in the press release as resembling an "old-fashioned hair dryer".