Computational and Cognitive Neuroscience,
will speak on
Dept. Zoology & Neurobiology,
Ruhr-University Bochum, Germany
Perception of Travel Distance from Optic FlowAbstract:
Effective navigation through an environment requires knowledge about the direction of motion and the distance traveled. Humans can use optic flow do estimate the direction of self-motion. But on its own, i.e., without additional cues about the 3D structure of the environment, optic flow is ambiguous with regard to travel distance. In my talk I will pursue the question under which conditions humans can perceive travel distance from optic flow. First, I show that humans are able to discriminate and reproduce distances of two visually simulated self motions when they are performed in the same simulated environment. In this case, two visually simulated motions can be discriminated under the assumption that the depth structure of the environment is not changed between the motions. Subjects are asked to discriminate distances of two visual simulated self motions of different velocity and duration in four different environments. These environments varied in the depth cues given to the subjects. We found that discrimination was possible in all environments. This suggests that motion parallax, which was the only depth cue available in all environments, is sufficient. In a second set of experiments, we modified the visual information about the structure of the environment between the two motions, thereby altering the speed distribution in the flow without altering the travel distance. This included variations of the height above ground and variations of the visibility range. When the height of the observer above the ground was changed without the subject noticing the change, errors occured that could be predicted by the change in flow speed due to the modification of height. When the height was altered visibly and the subject watched the height change, thw errors were strongly reduced. The results of the second set of experiments suggest that humans can transfer distance estimates from visual flow across changes in the visible structure of the environment. In a third experiment, we asked subjects to actively reproduce the depth of a given movement by driving the simulator with a joystick. Subjects were well able to do the task. Their behavior suggested that they remembered and reproduced the velocity profile of the seen movement.
The lecture will take place in the Lecture Hall, Room 203, 44 Cummington St.
on Wednesday, August 8, 2001
at 4:00 pm
Hosted by the Brain and Vision Research Laboratory