| Summary: | 碩士 === 國立臺灣大學 === 心理學系 === 82 === The 'form' could serve as an entity for perceptual
inference. However, it should be conceived as a composite
of monocular and binocular components with their
interactive characteristics. In experiment 1, we demonstrate
that monocularly visible form can not affect the
matching process of dots inscribed in a stereo pair.
Experiment 2 further demonstrates that different
interpretations of the same binocularly fused stuff will
affect perceived depth of the inscribed dots. For one percept
of the binocular form, the inscribed dots sit on the zero
disparity plane. However, for another percept through the
same binocular fusion materials, the inscribed dots are
captured to the positive disparity plane. Experiment 3 shows
that perceived transparency is solely determined by the
spatial relationship between the monocular and the
binocular forms. Theories of mutiple matching and lightness
difference are not sufficient for tackling with mechanisms
behind such perceived transparency. It is then proposed
that 'form' as a composite of monocular and binocular
components could be conceived as an emergent
representation following modular computation. Such a
knowledge representation may exert high-level feedback
to modulate stereopsis and perceived transparency.
|
|---|
