Can Spatial Biases Be Eliminated Through Learning?

Zhi Li,Yongchun Cai,Ci Wang

DOI: https://doi.org/10.1167/16.12.287

2016-01-01

Journal of Vision

Abstract:The perceived spatial layout is not veridical (or accurate). It has been shown that perceived slant is exaggerated, perceived in-depth distance is foreshortened, and even perceived 2D orientation/angle is distorted. What causes these spatial biases? Are they simply due to the lack of experience? The present study examined whether the spatial biases could be eliminated by experience. A simple form of spatial bias, the horizontal vertical illusion (HVI), was studied. Experiment 1 used a size estimation task, in which the participants judged whether the actual size of a horizontal (or vertical) line shown on the center of a CRT screen was longer or shorter than that indicated by a size label (i.e. "4 cm", "6 cm", or "8 cm") shown on the upper left corner of the screen. The results showed that the perceived size of 2D lines was substantially underestimated, and the underestimation was stronger for the horizontal lines. This size-anisotropy in perceived 2D lines well predicted the magnitude of HVIs obtained from a size comparison task with the same group of participants, which indicated that the HVI is due to perceptual biases in perceived size. Experiment 2 examined whether the HVI can be eliminated by training (i.e. by providing feedback) in the size estimation task. The results showed that, after training, participants' performance in the size estimation became pretty accurate; however, the HVIs obtained from the size comparison task remained intact. These results suggested that the perceptual biases in perceived 2D size were not affected by the training in the size estimation task, and the fact that the size estimation became more accurate after training was probably due to cognitive learning rather than perceptual learning. Meeting abstract presented at VSS 2016