In summary, they show that presenting volunteers with words such as "walker", which express human motion, increases the chances of them reporting the presence of human figures in noisy displays, not only when a target figure is presented (masked in noise) but even when no figure is actually presented. This indicates a potentially powerful role for language in shaping our interpretation of the visual world. Flecken and her colleagues are now analysing brain-imaging data that will help them to localise the interface between language and vision.

Monique Flecken spoke to BNR news radio about this research on 20 September.