The role ofoptical velocity in the control ofstance

Two experiments were designed to investigate the use of optical flow for postural control as a function of its velocity, geometry, and retinal placement. In the first experiment, subjects were exposed to a room that moved at a velocity below the reported threshold for detection of point­ light motion. In the second, the room was moved so that optical velocities were above this threshold. Compensatory sway was assessed for both lamellar and radial flow to central and peripheral areas of the retina. Compensatory sway was elicited with optical velocities below the point-light mo­ tion threshold, suggesting that this threshold is not relevant to the detection of egomotion. The results also indicate that the retinal periphery does not detect posture-relevant information from flow fields with a radial dynamic structure. Sincethe higher velocities used in the second experi­ ment exceeded those generated by natural postural instabilities, it was concluded that radial flow in the retinal periphery is not used as a source of information for normal postural mainte­ nance, and therefore that flow structure is an important factor in the detection and control of egomotion. Stoffregen (1985) recently reported a series of experi­ ments on the optical control of stance. These studies pur­ ported to demonstrate the importance of structure in op­ tical flow fields as a determinant of their usefulness as sources of information for postural control. It was ob­ served that the slight forward and back swaying that characterizes natural postural instabilities generates global optical flow, and that the structure of this flow in the am­ bient optic array varies depending on its proximity to the anterior/posterior line of motion. With any linear motion, if we look in the direction of motion, optical flow expands radially outward from the point toward which we are moving and sweeps laterally past us, converging behind. If we look at different parts of this overall flow, we find different local structures. Near the line of motion, the pattern of flow is almost ex­ clusively radial, whereas at those points where the flow sweeps past the observer, the lines of flow have become nearly parallel, much like the lines oflongitude on a globe, which converge at the poles but are parallel at the equator. Studies that have ignored this variation in flow struc­ ture have concluded that the periphery of the retina is dominant for the pickup of optical information specify­ ing egomotion (Brandt, Dichgans, & Koenig, 1973), but Stoffregen's (1985) results suggested that this was true only when the periphery was presented with flow that had a particular structure, the nearly parallel arrangement These experiments were carried out at Comel1 University in the labora­ tory of Eleanor J. Gibson. Her generous support is gratefully ac­ knowledged. The data were reported in a talk at the Third International Conference on Event Perception and Action, Uppsala, Sweden, June 1985. Thanks are extended to Gary Riccio, James Cutting, Mark Schmuckler, and Rik Warren, who provided helpful comments on the manuscript. Requests for reprints should be sent to Thomas A. Stoff­ regen, who is now at AAMRLIHEF, Wright-Patterson Air Force Base, OH 45433. described above (referred to as lamellar by Koenderink & van Doorn, 1981). The retinal periphery appears to be unable to detect posture-relevant information in radial flow patterns. Central retina shows a modest ability to detect such information from either radial or lamellar flow. A plausible conclusion is that flow structure is at least as important as retinal location in the control of stance. However, there is another explanation for the observed results. Variations in the optical velocity of points in difrerent parts of the array might account for the failure of peripheral retina to detect information for postural con­ trol from radial flow. 1 In an evenly cluttered environment, optical displacements caused by egomotion have the lowest magnitude, or velocity, near the focus of expan­ sion. 2 Optical velocities increase with increasing eccen­ tricity from the line of motion, until they reach a peak