When students reason during school science, they often refer to conceptions that are derived from out-of-school experiences and are poor proxies for science orthodoxy. However, for some areas of science, these conceptions represent only a proportion of students' full conceptual knowledge, for tacit understanding exists that is superior to the understanding displayed when reasoning. Noting that tacit understanding is engaged when events are judged as natural or non-natural, the paper is concerned with software that: (a) requires direction and speed of falling objects to be predicted, i.e. a typical science reasoning task that engages conceptual knowledge; (b) presents simulations of predicted motion in the expectation that its naturalness or non-naturalness would be recognised. An evaluation study is reported where children aged 8-12 years worked with the software in contexts that typify computer use in classrooms, i.e. individually under adult guidance ("n" ?=? 44 children) or in pairs with a classmate (n ?=? 48 children). They were observed while they did this. Reasoning about object fall was assessed via change from individual pre-tests administered prior to software usage to individual post-tests administered a few weeks afterwards. The children who worked with the software showed greater pre- to post-test gains in conceptual understanding than control children ("n" ?=? 47 children), who lacked software experience. The gains were especially marked for the children who worked in pairs. The approach taken is contrasted with traditional approaches to conceptual change in school science, e.g. strategies that rely upon classroom experiments. (Contains 1 note, 1 figure, and 2 tables.)