The finding is considered significant since conventional wisdom, and previous research using static images, had suggested that the inability to do this was almost a defining feature of autism.
The study, led by Elisa Back, compared two groups of autistic children with two control groups of non-autistic children. Researchers conducted two experiments in which the participants looked at a series of facial expressions on a laptop screen. In the facial images used, the eyes and mouth were either "freeze-framed" in a neutral expression, or animated and expressive. By showing a sequence of different combinations, they were able to gauge which aspects of the face were used by the autistic children to "read" someone's mental state and how successful they were.
In the second experiment, the 18 autistic children involved were as successful as non-autistic children in interpreting mental states, whether they saw the eyes in isolation or in the context of the whole face.