Visual perception of spatial relations in depth

Abstract

The visual perception of spatial relations of two objects was investigated in a series of experiments. We examined spatial and contextual parameters. The effect of spatial parameters was investigated with various two-dimensional tasks: an exocentric pointing task, a parallelity task and a collinearity task. Whereas spatial parameters like relative distance and visual angle influenced the settings of all observers in a similar way, there were differences between tasks in their dependence on different parameters. For example, whereas the settings of the other tasks were dependent on the relative distance the settings of the parallelity task were not. This can be explained by different task-demands that are specific for each of these tasks. In the exocentric pointing task an observer has to direct a pointer, with a remote control, towards a target. We expanded the exocentric pointing task to a three dimensional version in which the height was also varied. Therefore, we had two dependent variables: the deviations in the horizontal plane (slant) and in the vertical plane (tilt). With this extension of the task, we could conclude that visual space is anisotropic since in contrast to the slant the tilt was not dependent on the relative distance. Another three-dimensional task that has been used is the ball-in-plane task, a task in which the observer has to hang a ball in a plane defined by three other balls by adjusting it in height. We found settings that were best described as concave settings, which is in agreement with most conclusions of the work described above. Furthermore, we investigated the effects of context on the 3D exocentric pointing task. We tested whether the settings of the observer were dependent on an egocentric reference like frontoparallelity or an allocentric reference like parallelity to a wall. It turns out that people differ in the references they use to do the task. However, in another experiment, we concluded that a reference like one's own body position is impossible to ignore. Thus, it seems that cues that are really prominently present will be used by all observers, whereas for less prominent cues people can choose to use them or not. In conclusion, spatial parameters seem to affect all observers in a similar way, whereas contextual parameters can affect observers differently.