Mental models and other misconceptions in children’s understanding of the earth

127 six- and seven-year-old children

UK children aged 6-7, randomly given either the original Vosniadou & Brewer drawing task or a revised version with clarified instructions and questions.

A methodological replication-and-critique, and the right kind: it does not argue about Vosniadou & Brewer's coding, it manipulates the INSTRUMENT and shows the construct moves. Children received either the original protocol ("Can you draw a picture of the earth?", "Show me where the people live") or a version in which the ambiguity of perspective and scale was removed. The companion study in this programme (Nobes & Panagiotaki 2007, British Journal of Psychology) administered the ORIGINAL instructions to ADULTS, who also drew flat, hollow and dual earths and reported afterwards that the instructions were ambiguous and confusing - which is close to a decisive demonstration that the naive drawings are semantic and pragmatic errors rather than conceptual ones, since the adults demonstrably know the earth is a sphere. Read at abstract depth.

With the revised, disambiguated task, children "gave substantially more indication of having scientific understanding and less of having naive mental models" than with the original task - i.e. the synthetic models (flattened sphere, hollow sphere, dual earth) are largely METHODOLOGICAL ARTEFACTS of the elicitation procedure. The broader pattern across this research programme: children look like they hold naive mental models when asked to draw or model the earth and answer open, repeated questions, and do not when asked to choose among 3D models or pictures and answer forced-choice questions - and under the latter method their representations are either scientific or INCONSISTENT, not organised into coherent alternative theories. The authors also attribute part of the divergence from the 1992 results to cohort and cultural effects. FOR THIS ARCHIVE THE LESSON GENERALISES BEYOND ASTRONOMY: the single most influential demonstration that instruction produces new wrong theories rather than knowledge turns out to be partly an instrument effect - which is a direct warning about every bespoke misconception inventory in the conceptual-change literature, all of which are built and scored by the same people who designed the instruction.

Not applicable.

This IS the replication result. It does not overturn the finding that young children say non-scientific things about the earth; it overturns the inference that those sayings reveal coherent synthetic mental models. Vosniadou and colleagues have contested the interpretation, so the dispute is live rather than settled.