In an attempt to tap underlying competencies, the ability of children
with autism to comprehend pretend acts carried out by an experimenter
was compared with that shown by controls. These acts consisted of the
pouring of a pretend substance from an appropriate container onto a ta
rget figure. There was no significant difference in the groups' abilit
y to identify the pretend substance involved, to predict the pretend o
utcome of the actions, or to reflect on the pretend nature of the epis
odes. However, the performance of controls on these latter two tasks w
as surprisingly poor, limiting the implications that might be drawn fr
om the observed absence of group differences in comprehension abilitie
s.