Oakyoon Cha

The background is remapped across saccades

Physiological studies have found that neurons prepare for impending eye movements, showing anticipatory responses to stimuli presented at the location of the post-saccadic receptive fields (RFs) (Wurtz in Vis Res 48:2070–2089, 2008). These studies proposed that visual neurons with shifting RFs prepared for the stimuli they would process after an impending saccade. Additionally, psychophysical studies have shown behavioral consequences of those anticipatory responses, including the transfer of aftereffects (Melcher in Nat Neurosci 10:903–907, 2007) and the remapping of attention (Rolfs et al. in Nat Neurosci 14:252–258, 2011). As the physiological studies proposed, the shifting RF mechanism explains the transfer of aftereffects. Recently, a new mechanism based on activation transfer via a saliency map was proposed, which accounted for the remapping of attention (Cavanagh et al. in Trends Cogn Sci 14:147–153, 2010). We hypothesized that there would be different aspects of the remapping corresponding to these different neural mechanisms. This study found that the information in the background was remapped to a similar extent as the figure, provided that the visual context remained stable. We manipulated the status of the figure and the ground in the saliency map and showed that the manipulation modulated the remapping of the figure and the ground in different ways. These results suggest that the visual system has an ability to remap the background as well as the figure, but lacks the ability to modulate the remapping of the background based on the visual context, and that different neural mechanisms might work together to maintain visual stability across saccades.

Characteristics of the filled-in surface at the blind spot

Our visual system can restore information missing within the portion of the retinal image corresponding to the blind spot where the optic nerve exits the eye. Previous studies of the properties of filled-in surfaces at the blind spot have found similarities and dissimilarities between filled-in and real surfaces and have therefore not provided a consistent view of the characteristics of the filled-in surface. First, we investigated whether filling-in utilizes a contour integration mechanism. Gratings with collinear lines filled in the blind spot more effectively than those both with orthogonal lines and without any line, suggesting that collinear facilitation underlies the filling-in of the blind spot. Second, the dynamics of binocular rivalry was examined by comparing the dominance duration distributions of filled-in and real surfaces. The results indicated that the strength of the filled-in surface was attenuated compared to that of the real surface during rivalry. Lastly, we tested whether travelling waves of dominance in rivalry could occur at the blind spot. The travelling waves could propagate through a hole only at the blind spot, suggesting that the filled-in surface helps perceptual waves to travel across the blind spot. These results suggest that the filled-in surface shares a common mechanism via a horizontal connection but that it has weak strength to suppress the opposite eye during binocular viewing.

Perceived average orientation reflects effective gist of the surface.

The human ability to represent ensemble visual information, such as average orientation and size, has been suggested as the foundation of gist perception. To effectively summarize different groups of objects into the gist of a scene, observers should form ensembles separately for different groups, even when objects have similar visual features across groups. We hypothesized that the visual system utilizes perceptual groups characterized by spatial configuration and represents separate ensembles for different groups. Therefore, participants could not integrate ensembles of different perceptual groups on a task basis. We asked participants to determine the average orientation of visual elements comprising a surface with a contour situated inside. Although participants were asked to estimate the average orientation of all the elements, they ignored orientation signals embedded in the contour. This constraint may help the visual system to keep the visual features of occluding objects separate from those of the occluded objects.

Composite binocular perception from dichoptic stimulus arrays with similar ensemble information

We view the world through laterally displaced eyes that generate two streams of image signals differing slightly in their perspectives of the visual scene. The brain derives three-dimensional structures from these two image streams by establishing binocular matches and computing image disparities between the two eyes’ views. Since the binocular matching problem can have multiple, alternative solutions, vision relies on several strategies to determine the most probable matches. The current study investigated whether the visual system might utilize regularities among neighbouring features (feature ensembles) when confronting this problem. We hypothesized that binocular perception with unlikely, anomalous ensembles would indicate unsuccessful binocular matches. We made dichoptic stimulus arrays of coloured circles and manipulated the colour similarity of stimulus items to produce probable or unusual ensembles when superimposed. Using binocular rivalry as a proxy index, we found that composite perception of dichoptic arrays was more stable when the stimulus items shared similar colours, and that unusual ensembles induced binocular rivalry. Our results suggest that binocular ensembles can be utilized to detect unsuccessful binocular matches, thus uncovering a potentially useful supplemental strategy for identifying binocular matches when viewing potentially confusing visual scenes containing redundant visual features.

Dissimilarity between feature ensembles triggers binocular rivalry without competing local features

binocular rivalry without competing local features Oakyoon Cha1, Randolph Blake2, & Sang Chul Chong1,3 1Graduate Program in Cognitive Science, Yonsei University; 2Department of Psychology and Vanderbilt Vision Research Center, Vanderbilt University; 3Department of Psychology, Yonsei University #56 4060 Experiment 2: Does binocular conflict (i.e., overlap) influence rivalry dynamics of unitary percepts?

Relational information decays faster than object features in visual working memory.

Visual working memory (VWM) actively maintains visual information for a short period. While some researchers suggested that VWM representation consisted of slots storing bound objects (e.g., Luck et al., 2001), other researchers found that degradation of relational information decreased VWM capacity (e.g., Wilken et al., 2004). Recently, Brady et al. (2013) found that different features of objects decayed at different rates in long-term visual memory. However, the same hypothesis has not been tested in VWM. We investigated whether relational information among items decayed faster than object features in VWM. To test this hypothesis, we used a change detection task. We varied the retention intervals (short: 150 ms, long: 750 ms) and used different locations between test and memory arrays. The memory array was presented in the parafovea and the test array was presented around the fixation to prevent participants from utilizing sensory memory. There were three types of changes in the test array. In the color change condition, the color of one item was changed. Because colors of individual objects and relationship between colors were preserved, participants could use both object features and relational information. In the location change condition, relative location of one item was changed. Note that absolute locations of all items in test array were changed from those of the memory array because of the different locations between the test and memory arrays. Thus, only relational information was available in the location change condition. Finally, in the both change condition, both relative location and color of one item were changed. Change detection performance was similar at the short retention interval across the conditions. However, at the long retention interval, change detection performance of the location change condition decreased significantly more than the other conditions. These results suggest that relational information decays faster than object features in VWM. Meeting abstract presented at VSS 2015.