READING BACKWARD - PERCEIVED DIRECTION OF MOVEMENT IN CONTRACTING ANDEXPANDING OPTICAL-FLOW FIELDS

Two goals were pursued in an investigation of possible visual sources for directionality judgments of ego-motion. First, J. J. Gibson's (195 0) global radial outflow hypothesis was contrasted with a simple extra polation strategy Second, backing-direction judgments capitalizing on the informational equivalence of global radial outflow created during forward ego-motion and global radial inflow during backward ego-motion were explored. In comparing the accuracy of heading and backing judgm ents, ne rr insights about global flow and extrapolation strategies we re found. Consistent with the hypothesis of an extrapolation strategy, Experiment 1 demonstrated that backing judgments were more accurate t han heading judgments when linear observer motion was simulated by mea ns of a point-light flow field. In this case, accuracy was higher with two-point-light displays (extrapolation) than with more complex displ ays (global flow), Experiment 2 showed that in cases where extrapolati on was not possible, such as circular motion, no advantage of backing judgments could be found and judgments were generally less accurate. W e conclude that, whenever possible, observers use extrapolation to det ermine their heading/backing. Only when global flow is the only good s ource of information do they rely on it.