透明與不透明錯覺輪廓遮蔽面對於整體運動知覺的影響

The Effect of Transparent and Opaque Illusory Contour Occluders on the Motion Interpretation

根據中層次視覺理論觀點，面訊息對於整體運動知覺之處理具重要影響力，已獲許多支持證據。本研究接受此一觀點，沿用McDermott、Weiss及Adelson（2001）的實驗典範，針對遮蔽面線索之有助於形成一致運動知覺，進一步就錯覺輪廓遮蔽面及其透明性之效應加以探討。實驗一與實驗二分別以不同方式產生錯覺輪廓，結果顯示具錯覺輪廓遮蔽面可提升一致運動知覺比率，與真實輪廓遮蔽面具類似的效果，而透明錯覺輪廓所得效應低於不透明錯覺輪廓。實驗三至實驗五進一步將錯覺輪廓所得效應與其他類型輪廓作比較，並分別以不同方式操弄透明度，結果發現透明錯覺輪廓所得效應低於不透明錯覺輪廓得重覆驗證，顯示錯覺輪廓之圖形結構因素所產生的透明感，具有穩定的影響效果，然而實驗四之動態操弄透明度，只得到較小的效果，而實驗三、五以靜態深度操弄透明度，則透明性降低一致運動的假設未獲支持。對於真實輪廓操弄透明度，亦未呈現明顯的效應。綜合五個實驗的結果，初步支持中層次視覺理論觀點，然而面透明性訊息在不同情況下對於運動知覺產生不同的影響，則有待進一步探討。

Based on the theories of midlevel or mediumlevel vision, the information about visual surfaces can have critical effects on the motion interpretation. Many of the evidences have revealed to support this theoretical point. Following such aspect, the purpose of this study is to explore how the attributes of the surface affect the motion interpretation. The major focus is to test whether the illusory contour, instead of real contour, being the occluders can also have similar effect on the motion interpretation. In addition to this test, the study on how transparency of the occluding surfaces influence motion interpretation is also performed. The paradigm developed by McDermott, Weiss, and Adelson (2001) was used in the present study. In experiments one and two, two types of illusory contour were designed as occluders. The results showed that illusory contour occluders affect motion in a similar manner as that of real contour occluders. And the effect was smaller for the transparent illusory contour as compared to the opaque illusory contour. In the subsequent three experiments, we designed various types of contour and manipulated transparency by different methods for comparison. It was found that the smaller effect of transparent illusory contour, when compared to opaque illusory contour, could be replicated in all three experiments. However, in the condition of real contour, the effect of transparency was not shown. Also, manipulating the transparency by different methods showed different effects in these three experiments. The most stable effect was obtained in manipulating the figural configurations to induce transparency in the illusory contour. Transparency induced by dynamic motion cue only showed a minor effect. Manipulating transparency by static binocular depth did not show any significant effect. From the results of all five experiments, the important role of the surface formation in motion processing which was predicted by the midlevel or medium-level vision theories was partially confirmed. On the other hand, not to be covered by these theories, that transparency of the surface can influence motion interpretation only in some of the conditions was observed in this study. Further research will be needed to reveal more details of the issue.