Ninety-one children aged between 2;10 and 5;0 participated in a training study of false belief. Children were assigned to either an explanation condition, a practice condition or a control condition where children heard two stories unrelated to false belief. Children's eye movements in anticipation of the protagonist reappearing were monitored at pre-test.Only the explanation condition led to improvements in judgement and justification of a protagonist's future action based on false belief. Children who looked in anticipation to where the protagonist thought the object was at pre-test were more likely to give a correct judgement at post-test than those who did not. Those children in the explanation group who gave a correct judgement at pre-test were more likely to give an appropriate justification at post-test than those who did not.Three main conclusions are drawn: (1) providing explanation about the underlying principles of a task is more likely to lead to improvements in performance than merely informing children of whether their response is correct; (2) the nature of improvements in performance will depend on the level of knowledge of the child at pre-test; (3) training will only be beneficial for those children who demonstrate evidence of implicit understanding.

We presented children aged 6, 8, and 10 years with a video and then an audio tape about a dog named Mick. Some information was repeated in the two sources and some was unique to one source. We examined: (a) children's hit rate for remembering whether events occurred and their tendency to make false alarms, (b) their memory for the context in which events occurred (source monitoring), (c) their certainty about hits, false alarms, and source, and (d) whether working memory and inhibition were related to hits, false alarms, and source monitoring. The certainty ratings revealed deficits in children's understanding of when they had erred on source questions and of when they had made false alarms. In addition, inhibitory ability accounted for unique variance in the ability to avoid false alarms and in some kinds of source monitoring but not hits. In contrast, working memory tended to correlate with all forms of memory including hits.