Abstract
(1) Much developmental evidence based on the so-called standard or elicited false belief task shows that when asked to predict where an agent with a false belief about an objects location will look for it, children who know the location of the object fail until they are well into their fifth year. (2) However, several more recent experiments based on different paradigms, including the violation-of-expectation paradigm, anticipatory looking, helping and referential communication, also show that preverbal human infants seem to reliably represent an agents false beliefs. The puzzle is: how to reconcile (1) and (2)? Until recently there were two main strategies for solving the puzzle. One strategy is to take the data on preverbal human infants at face value and show why it is so hard for 3-year-olds to pass the standard false belief task. The other strategy is to offer low-level explanations for the data on preverbal human infants and deny that they are able to represent anothers false beliefs. A third strategy has recently emerged, based on a two-systems approach to belief-ascription. On behalf of the first strategy, I will argue that there is decisive evidence against the second strategy and that it is hard for the third strategy to get off the ground.