Hitting what one wants to hit and missing what one wants to miss.

When an observer gazes directly at a rigid spherical object moving at constant speed along a line directed at the head, both monocular and binocular retinal image correlates of time to collision (TTC) are available provided that the object is not too small. The monocular correlate is not available for very small objects and is invalid for rotating aspherical objects, while the binocular correlate is available only when the ratio (closing speed)/(distance) is sufficiently large. Both cues are maximally effective in the central visual field so it is helpful to foveate potential collision hazards. On the other hand, in the special case of prolonged periods of driving along a straight empty road it is important to vary the direction of gaze rather than continuously gazing straight ahead so as to avoid the local adaptation to retinal image expansion that can cause errors in judging TTC when only monocular information is available. A more benign effect of self-motion is a long-distance interaction between the TTC signal generated by the approaching object and the expanding flow pattern caused by self-motion. This interaction creates a margin of safety. We also discuss eye movement strategies in executing the following two tasks: estimating the direction of self-motion; hitting a cricket ball.