Abstract
A change in radius of a stationary object is signalled to a stationary observer by a change in retinal size. Such a change is readily detected. If the object is approaching the observer then detection is more difficult – change in retinal size is no longer sufficient. The observer needs to pick up a change in the relationship between the monocular and binocular cues to motion-in-depth. If the observer is moving and the object is embedded in a stationary scene then an additional solution is available. It has recently been proposed (e.g., Rushton & Warren, 2005, Current Biology; Warren & Rushton, 2009, Current Biology) that during self-movement the brain uses its sensitivity to optic flow to identify and subtract out components of global retinal motion that are due to self-movement (“flow parsing”). This isolates components of retinal motion that are due to movement of objects relative to the scene. If the brain performs flow parsing, detection of a change in object radius should be easier when the observer approaches the object rather than when the object approaches the observer. Across a series of experiments we explored this prediction. Observers viewed a scene consisting of 25 wireframe objects arranged in a volume of 60 cm depth, centred 90 cm away. In the first experiment a central sphere moved towards the observer. Observers had to indicate whether or not the approaching object was changing size. On 50% of the trials the sphere changed in radius during approach, on the other 50% it did not. Performance (group average, N = 5) was barely above chance. In a second experiment the central sphere moved (and changed radius) in the same way but now the whole scene approached the observer. Performance was above 90% correct. We conclude that observers can detect a change in size during self-movement and that performance is compatible with the involvement of a flow-parsing process.