Abstract
During visual fixation, unless the head is artificially constrained, involuntary, small head movements significantly contribute to the motion of the retinal image (Skavenski et al., 1979). In this study, we investigated whether the resulting retinal image motion provides useful depth information in the form of motion parallax. Head movements in human observers were measured by means of a custom-developed high-resolution method, which uses a Phasespace motion capture system with 4 cameras (480 Hz) and a tightly-fitting helmet with 20 active optical markers. Preliminary measurements with the helmet mounted on a robotic manikin head have shown this method gives resolution higher than 2′ for head rotations and better than 1 mm for head translations. Recordings were made while subjects maintained prolonged steady fixation on targets located at distances of 0.5–3 m, while standing or sitting with their head unrestrained. The mean translation and rotation velocities measured in the experiments were 3.6 mm/s and 25′/s, respectively. Physical modeling of the head and eye enabled estimation of retinal image motion. Significant retinal image motion was found even under the assumption that eye movements perfectly compensated for head movements, so to yield complete stabilization of the fixated target. For example, during fixation on an object at 130 cm, retinal speeds larger than 1′/s were already found for objects located just 10 cm away from the fixation point. The speed of the retinal image increased monotonically with the distance of the object from the fixation point yielding velocities well above thresholds. Furthermore, retinal velocities increased even more if perfect stabilization of the fixated target was not assumed, as it is known to occur from previous studies. These results show that retinal image velocities caused by fixational head movements are within detectable levels for the visual system and may contribute to depth perception.
NSF (CCF-0720691) and NIH (R01 EY18363).