The ‘windows task’ as a measure of strategic deception in preschoolers and autistic subjects

Abstract

We gave 3‐ and 4‐year‐old children and autistic subjects (plus a group of Down's syndrome children of equivalent ability) a task which measured their capacity for strategic deception. This was a competitive game played between the subject and an experimenter in which the participants tried to win chocolates. In the training phase the subjects learned that it was in their interest to tell the experimenter, by pointing, to look into an empty box for the chocolate, although subjects did not know until after the search which box was empty. In the testing phase, the boxes had windows facing the child so he or she could now see which was the empty box. The 4‐year‐olds and the Down's children, but not the 3‐year‐olds and the autistic subjects, generally pointed to the empty box on the first trial. Moreover, the younger children and the autistic subjects frequently continued to point to the baited box for the full 20 test trials. We also found that the ability to apply the correct strategy on the ‘windows task’ was associated with success on a standard ‘false belief’ task and argue from this that both tasks may be difficult because they require subjects to inhibit the tendency of salient knowledge about object locations to overwrite knowledge of epistemic states.