Abstract
Action measures reflect the calibrated relationship between perception and action (Powers, 1973). There is evidence that egocentric distances are underestimated in normal environments even though people walk them accurately. One basis for this claim is that when people are asked to match a frontal extent with an egocentric one, they set the egocentric interval much too large. Li, Phillips and Durgin, (2011) conducted such matching experiments in both (panoramic) virtual (VR) and real outdoor environments. Similar matching errors were found in both environments, as if egocentric distances appeared compressed relative to frontal ones. In the present study we compared action measures (visually-directed walking) for egocentric and frontal intervals in VR and in an outdoor environment. Walking estimates of frontal distances were relatively accurate in VR, but walking estimates of egocentric distances were short. Geuss et al. (2011) have interpreted such a pattern of data as indicating that egocentric distances, but not frontal extents, are compressed in VR. However, the ratios of walking in the two conditions exactly correspond to the matched ratios found in the matching task both in VR and in an outdoor environment. Moreover, we found that walking measures overestimate frontal extents in outdoor environments (see also Philbeck et al., 2004). It seems that frontal intervals and egocentric intervals are both compressed in VR. Frontal intervals may be matched relatively accurately in VR by walking measures because the compression of VR approximately offsets the errors that are normally observed in real environments. Walking actions are calibrated during normal use, but walking is normally used to cover egocentric distances, not frontal ones. Because frontal intervals appear larger than egocentric intervals, it should be expected that walking out frontal intervals will produce proportionally greater estimates than walking out egocentric intervals even in VR.
Meeting abstract presented at VSS 2012