Abstract
Visual motion is a powerful signal that provides information about where we are heading, what is figure and what is ground or the material properties of objects in the world. Often visual motion is self-generated through active exploration of the world, e.g. by walking, moving our heads or by poking and kneading objects with our hands to see them wobble, sparkle or deform. Using a novel task we sought to understand how self-generated exploratory actions are tuned towards the task and visual properties of objects. Stimuli were stacked, in size decreasing, 2D rounded squares that varied in positional inertia (.25, .5, 1, 2, 4 where 0 implies all squares move at the same speed, and increasing values make lower layers move relatively slower) and number of layers (3, 4, 6, 10, 18). Fifteen participants explored a stimulus on a given trial by dragging it around on a tablet while we recorded the finger trajectory. Following the exploration, their task was to judge whether the stimulus appeared to be 2D or 3D and whether it looked rigid or soft. Each condition was repeated 5 times. The software was written by us in Java. There was a high agreement for 2D/3D judgments between all participants, except for 2 who where excluded from the subsequent analyses. Finding reveal significant effects of number of layers and positional inertia on overall relative frequency of 2D-, and rigid judgments, as well as significant interactions between the two stimulus parameters. There were also significant main effects of layer number and positional inertia on exploration time and distance as well as significant interactions between these two stimulus parameters. Moreover, inspecting the individual exploration paths reveals marked differences in how participants explore these novel stimuli (swirling vs linear exploratory finger motions). Finally, we relate exploration characteristics to perceptual judgements.