The 60 reception-class children involved in the study were split into an experimental group, which received adult help, and a control group, which did not. They were then given problem-solving tasks such as matching two-dimensional outlines to the faces of three-dimensional wooden blocks.
In both groups, some children who had initially failed to complete a task managed to do so without any help once it was placed in a meaningful context. When they were asked to balance a cuboid on three cones, most were unsuccessful.
But when asked to build a rocket from a number of blocks, including three cones, 46 per cent of those in the control group and 52 per cent in the experimental group solved the problem.
After completing a task, children in the experimental group received feedback, which helped them apply what they had learnt to new situations.
Those who had not managed to find a solution received graded levels of adult support. This ranged from changing the task to make it less abstract, to showing the child how to complete it.
The majority managed to complete the task after receiving the third level of help - a demonstration of the actions needed to solve the problem, using an equivalent but different set of blocks.
When presented with a task similar to one they had tackled, but set in a different context, 78 per cent of the experimental group found a solution without any adult help - compared to the 52 per cent who had managed to do so the first time round. The results of the control group, on the other hand, improved by less than two percentage points.
The researchers conclude that graded levels of adult support - or "scaffolding" - are more effective in helping children acquire basic geometrical concepts than simply giving them extra tasks to do.
"Scaffolding learning through meaningful tasks and adult interaction", by Penny Coltman, Dinara Petyaeva and Julia Anghileri, Faculty of Education, University of Cambridge. Early Years, Vol. 22, No 1, 2002