Congratulations to Dr. Marina Shpaner on
the successful defense of her doctoral thesis, "Neurophysiological
Investigations of Visual Object Completion Processes" on Friday, January
21st, 2011!
Abstract:
Research over the past several decades
provided exciting insights into the cortical underpinning of vision. While
early efforts concentrated on identifying and characterizing the distinctive
properties of neuronal populations, more recent investigations focus on how
separate channels of information are integrated. We examined the cortical dynamics
of object completion within the framework of interacting parallel pathways. The
“frame and fill” model advances fast and coarse scene analysis in the dorsal
visual stream as the foundation of more detailed object-related processing in
the ventral visual stream. Our first study assesses sensitivity to illusory
contour (IC) stimuli within the lateral occipital complex (LOC) as compared to
salient region (SR) stimuli, a stimulus class that lacks the classic bounding
contours of illusory shapes. Using high-density electrical mapping of visual
evoked potentials, we show that early LOC activity is substantially stronger to
IC than to SR stimuli, while later LOC activity is stronger to SR than to IC
stimuli. We further suggest that crude region based segmentation takes place in
the dorsal stream regions prior to contour interpolation. The second study more
closely examines contour interpolation in the context of complex visual scenes.
We use different levels of clutter to obscure the salience of the centrally
presented IC, thus taxing the initial scene delineation process. In support of
the “frame-and-fill” model, we show that high levels of clutter significantly
affect the early illusory contour sensitivity in the LOC and result in an
apparent switch to more effortful mode of contour interpolation. The final
study focuses on mechanisms of contour integration in the ventral stream. We
arbitrate between two previously advanced mechanisms of contour integration:
serial facilitative interactions between collinear cells in the primary visual
cortex or “coincidence detection” in higher order visual areas. We use
retinotopic presentations of gabor contours embedded in gabor noise in order to
assess the timecourse of earliest cortical activity. We report earliest contour-related
cortical responses in the higher-order visual areas and not in the primary
visual cortex, indicative of “coincidence detection.” Overall, the results of
our studies support the tenets of the “frame-and-fill” model of cortical
processing.