Item

Abstract

Purpose. We investigate the processing of visual information about rotations in three-dimensional space. Such rotations take place continuously when one is manipulating tools or objects with one's hands; both the accurate perception and the precise control of the orientation of an object are often of vital importance for a successful manoeuvre. This notion inspired us to introduce a new paradigm: we ask subjects to mimic the rotary movement of objects displayed on a computer-monitor by dynamically changing the orientation of a real object that they hold in their hand. Naturally, the feasibility of such an approach will have to be proved by the outcome of our experiments. Methods. The observer watches a polyhedral wire-frame object (computer graphics) that is rotating back and forth about a fixed axis in space. At the same time the observer is asked to manipulate a real object (a ball) so that it rotates in a similar manner. Afterwards, the recorded motion of the real object is analysed in detail. Results. We find that subjects can indeed carry out the task in a fairly accurate way, at least as far as the orientation of the rotation axis is concerned. The variability in the reproduced orientation of this axis is lower for rotations about (nearly) fronto-parallel axes than for rotations about slanted axes. The standard deviation (S.D.) of signed angular error in arbitrary directions varies from 5° to 15°, depending on slant and on subject. This S.D. is higher than the S.D. of 1-4° that is found when subjects are asked to rotate the real object about one of four pre-defined canonical axes (without a visual stimulus); we think that the latter S.D. reflects the baseline response accuracy of the motor system involved in manipulating the real object. We find significant systematic deviations from veridicality (veridicality in terms of the geometric model that we use to generate the stimuli). These biases can be described by an overall rotation of response space relative to stimulus space, combined with a reduction in the range of measured slants of the rotation axes relative to the stimulus range of slants of the rotation axes. Conclusions. We find that subjects are sensitive to visual information about the simulated three-dimensional rotation of a virtual object, and that they can use this information to mimic an object's motion by making hand movements in the real world. This result proves that it is feasible to use a motor action to investigate visual information processing.