非表示:
キーワード:
-
要旨:
The present study has a twofold purpose: to see whether observers can learn to discriminate curved shapes specified by motion, and to see how their performance is influenced by uninformative contextual stereo information. The stimulus was a random-dot display depicting an oscillating planar or curved surface whose central circular region was modulated to produce a quadratic surface that could depict either a 'saddle' or a 'hill'. The surrounding region was defined by motion and stereo cues. In contrast, the central region only provided motion information since stereo correspondence between dots on the two monocular images was not possible in that region. Consequently, stereo information only provided a context to the motion-specified shape. Moreover, motion information was identical in all contexts. Through feedback trials, observers learned to discriminate between the two shapes in a given stereo context. Following learning, observers were tested in the learned context and a new context. The results indicate that shape discrimination can improve through learning, and that contextual information can modulate task performance.
