Body-centered representations for visually-guided action emerge during early infancy.

The nature of the spatial representations that underlie simple visually guided actions early in life were investigated through the application of a 'double-step' saccade paradigm to 3- and 7-month-old infants. Saccades in the older infants, like those observed in adults, were executed within body-centered spatial coordinates that take into account the effects of intervening eye movements. In contrast, younger infants tended to respond according to the targets' retinocentric locations and did not incorporate the effects of displacements caused by previous saccades. These results indicate that contrary to prevailing views, body-centered representations for action are not present from birth but emerge, probably through experience, over the first few months of life.