Abstract The penny-hiding game is a deception game that occurs naturally in parent-child and child-child interaction. It involves minimal linguistic demands, and is lots of fun. Oswald and Ollendick (1989) employed it with subjects with autism and reported an impaired capacity for deception. They also found that this correlated with performance on both a false belief (“theory of mind”) test as well as various measures of social behaviour. The experiment reported here set out to replicate Oswald and Ollendick's important results, and then extend them by using a new technique for error analysis.
We succeeded in replicating the autism-specific deception impairment as well as the finding that deception capacity correlates highly with performance on a false belief test. In addition, the new analytic technique discriminated the group with autism from controls more clearly than the traditional index of deception. Specifically, subjects with autism, whilst fully capable of enjoying the game as a game of object occlusion (keeping things out of sight), failed to perceive the game as a game of information occlusion (keeping things out of mind), unlike normal children or subjects with a mental handicap of an equivalent or lower mental age. The dissociation in autism between occluding objects vs occluding information is discussed in relation to other research showing that subjects with autism are impaired in understanding the principle that “seeing leads to knowing”.