So far, Kuhl and her colleagues have already seen differences in brain activation in infants who are presented with language spoken either by a person in the room with them, or the same person, speaking the same script, on a television. "Once we know what the difference is between live and television exposure, we would like to take same measures from children with autism," Kuhl says. "They may be more engaged in a television set."

Using a less exact technology called electroencephalography, or EEG, her group has previously found that typical children respond to a classic mothering voice, while children with autism are far more interested when the same tones are produced by a computer. However, EEG has poor spatial resolution, making it difficult to determine where in the brain these differences arise. Now these kinds of studies can be performed in greater detail with MEG to better understand the brain areas involved. "Brain measures are going to be extremely potent biomarkers of autism," Kuhl says. Early diagnosis, before the first visible symptoms appear, may lead to early interventions.

"No one has done [MEGs on young children] in a systematic and vigorous way before," says Steven Stufflebeam, the director of clinical magnetoencephalography at the Martinos Center for Biomedical Imaging at Massachusetts General Hospital. "If they do it, they may discover something brand new that could revolutionize the neuroscience of kids. But it's a little hard now to predict what they're going to find."