Computational mechanisms of object constancy for visual recognition revealed by event-related potentials.

This study examined the functional organisation of the computational processes underlying orientation-dependent and orientation-invariant two-dimensional object recognition. Participants identified two previously memorised novel shapes at different image plane orientations while event-related potentials (ERPs) were recorded. A centro-parietal ERP component was found that peaked between 350 and 450 ms post-stimulus onset and whose amplitude was modulated by stimulus orientation only for objects showing an orientation effect in response times. These findings are consistent with a serial model of object recognition whereby object constancy is achieved in at least two successive steps: orientation-invariant feature extraction and orientation-dependent visuo-spatial transformation.