Abstract
Previous research has shown that peer collaboration can facilitate conceptual change in physics, as long as learners approach the exercise with differing conceptions about the issue at stake. Positive results have been obtained across a wide age range, with a broad selection of topics, and with contrasting modes of task presentation. Despite this, there are indications that differing conceptions is not the only variable significant to outcome and that task design may also be relevant. In this article, we report a study and consider this possibility in the context of 8- to 12-year-olds' conceptions of heating and cooling. Tasks were varied according to (a) whether their problem materials were or were not structured to permit critical testing and (b) whether they did or did not incorporate instructions to generate summary rules. The results showed the general superiority of tasks that both facilitated critical testing and required rules. The features did not operate additively, however, for there were indications that designs that deployed one feature in isolation were less helpful than designs that deployed none. Detailed analysis of the on-task dialogue has clarified why critical testing and rule generation were powerful in combination but not separately, showing that together the two features ensured progress at the level of theory and not merely empirical regularity. This not only is in complete contrast to the other designs but also helps to explain why the results were similar yet subtly different from those obtained in a comparable study whose topic area was floating and sinking.