Purchase this article with an account.
Hiroshi Ando; Visual learning in the spatial prediction of an approaching 3D object. Journal of Vision 2001;1(3):313. doi: https://doi.org/10.1167/1.3.313.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose. Visual prediction in a dynamic environment may play a crucial role for successful action control. Although a temporal aspect of visual prediction, i.e., time-to-contact, has been extensively studied, few have addressed the issue of the spatial prediction of dynamic events. The present study investigated the visual prediction of the future position of an approaching 3D object. In particular, this study examined whether spatial prediction improves with visual learning. Methods. The experiments measured the accuracy in predicting the future position of an approaching virtual object and compared the performance before and after a training session. In the first session, a stereoscopic spherical object was projected from a distant point towards an observer at a constant velocity (4 m/sec). The object moved 6m in two different approaching directions. The end position varied around the target, which was located 0.4m from the observer. The object disappeared before reaching the end point, and as soon as the object disappeared, the observer judged with a key press whether the object would eventually reach the left or right of the target. The training session was designed so that the object would appear at the end position immediately after subjects responded only for one of the approaching directions. The second session repeated the same procedure as the first one with no feedback. Results. The prediction accuracy and its confidence measure were estimated by fitting the logistic function to the position judgement data. The estimation results along with response time analysis indicated that the subjects could predict the future position of the object with an error of at most a few cm. Furthermore, the prediction performance improved after the training only for the trained direction. Conclusions. The results suggest that the visual system has an ability to predict the future position of a moving object and that prediction performance improves with visual training.
This PDF is available to Subscribers Only