Pii: S0010-0277(99)00046-3

The work reported provides an information processing account of young children's performance on the Smarties task (Perner, J., Leekam, S. R., & Wimmer, H. (1987). Three-year-olds' dif®culty with false belief: the case for a conceptual de®cit. British Journal of Developmental Psychology, 5, 125±137.). In this task, a 3-year-old is shown a Smarties tube and asked about the supposed contents. The true contents, pencils, is then revealed, and the majority of 3-year-olds cannot recall their initial belief that the tube contained Smarties. The theoretical analysis, based on the Headed Records framework (Morton, J., Hammersley, R. H., & Bekerian, D. A. (1985). Headed records: a model for memory and its failures. Cognition, 20, 123.), focuses on the computational conditions that are required to resolve the Smarties task; on the possible limitations in the developing memory system that may lead to a computational breakdown; and on ways of bypassing such limitations to ensure correct resolution. The design, motivated by this analysis, is a variation on Perner's Smarties task. Instead of revealing the tube's contents immediately after establishing the child's beliefs about it, these contents were then transferred to a bag and a (false) belief about the bag's contents established. Only then were the true contents of the bag revealed. The same procedure (different contents) was carried out a week later. As predicted children's performance was better (a) in the `tube' condition; and (b) on the second test. Consistent with the proposed analysis, the data show that when the computational demands imposed by the original task are reduced, young children can and do remember what they had thought about the contents of the tube even after its true contents are revealed. q 1999 Elsevier Science B.V. All rights reserved.