Session Detail

It has shown that the contextual information influenced judgements of the target facial expression. In this study, we systematically investigated how the surrounding emotional faces influenced expression perception by using categorization, discrimination, and rating tasks. In the categorization task, the participants had to categorize the target faces as happy or angry while the target facial expressions morphed to five levels varying between angry and happy faces. We also measured the response time in the categorization task. In the discrimination task, we required participants to discriminate a facial expression (either happy or angry) from a neutral face. The target morphed between happy and neutral faces or angry and neutral faces. In the rating task, the participants needed to rate the intensity of the angry and happy target faces. In all of the experiments, the targets were surrounded by five different types of face conditions (happy, angry, neutral, inverted neutral, and absent [baseline] faces). Despite the fact that the results showed that the surrounding conditions did not affect categorization, discrimination, and intensity judgement, they had a main effect on the reaction times during the categorization task. Hence, participants had the fastest reaction times when identifying target faces with happy surrounding conditions.

Accurately identifying other’s emotional states is crucial for fluent social interaction. The Cambridge Mindreading (CAM) Face Battery is widely used to assess one’s emotional recognition ability with 24 sets of video-taped subtle emotional expressions. However, the video stimulus is limited to Caucasian population and include emotional expressions uncommon in Asian culture. Thus, the purpose of this study is to develop and validate a Korean version of CAM(Yonsei-CAM). This study incorporated 18 commonly used emotional expressions in Korea which were chosen on the base of 80% agreement from 11 reviewers. The selected expressions were then recorded with 56 female and male Korean actors aged between 20 to 40. After all the clips had been received, two independent observers compliant to the observer reliability selected 5 seconds from each clip that best represent the corresponding emotional expression. The final clips will be selected upon three criteria after conducting an assessment with 120 college students: the accuracy rate of tasks, the reported degree and the smoothness of the emotions. Y-CAM can contribute in diverse fields of academia as the face stimuli can be utilized in studies with Asian population.

We examined the influence of head orientation on perceived gaze direction by using a categorization task and adjustment of an on-screen pointer. In both tasks, we found that the repulsive influence of head orientation on perceived gaze direction was greater when only the model’s farther eye was visible compared to when only the nearer eye was visible. When both eyes were visible, the two tasks revealed a different pattern of results suggesting a flexible use of information from two eyes depending on the task. We also analyzed the relative position of iris within the eye-opening in the stimulus. The analysis revealed a greater repulsive influence of head orientation for the model’s farther eye compared to nearer eye, which is consistent with the perceptual performance. The image analysis also revealed that the repulsive influence of head orientation on image eye region information was much greater than previously inferred based on gaze judgement performance with eye-only images (Otsuka et al., 2014; 2015). The current results suggest that observers in our previous studies used residual head orientation cues in the shape and/or relative size of the two eyes to counteract the repulsive influence of head orientation, causing an underestimation of the repulsive effect.

Building a culturally-calibrated face database with variations in identity, expression, pose, and lighting, can meet the demand of a multitude of face research. The present study represents an extension of our previous effort in this regard (Shyi, Huang, & Yeh, 2013), where subsequent studies have exposed its limitations in fulfilling the requirements placed by experimental manipulations. 6,600+ face photos, varying in terms of expression, pose, lighting, and gaze direction were taken from a group of 36 female and male models. About half of them (3,300+) were rated by another group of 120 participants to ensure that each face image was rated by at least 10 different participants. We then categorize each images into one of the seven basic emotions in terms of intensity and polarity. In order to compare the current and the previous databases, we conducted analyses to examine how pose, intensity and direction of lighting may affect the ratings and entropy values associated with each image. The overall results from these analyses show patterns remarkably similar to the previous database. Taken together, we consider the current database and the previous database are empirically equivalent, and can be linked into a larger and more useful source for face research.

Autism spectrum disorders(ASD) are characterized by deficits in social interaction and communication, as well as restricted behaviors and interests. A body of literature shows that ASDs have impairments in recognition of facial emotion which may contribute to social impariment. Most studies use faces presented alone without any context as stimuli, suggesting strong and automatic influence of context on face recognition. Consequently, prior studies for ASDs may be limited in the sense that they lack potentially important contextual cues and may not fully capture the difficulties experienced in everyday life. This study investigated quantitative change in emotional intensity ratings with the addition of contextual information in children with ASDs. Participants in the current study were 19 children with ASD. All participants were asked to assess the emotional intensity of a single emotion(happy/anger) from images presented under two conditions(context-free and context embedded). The results showed there is no significant difference in the addition of contextual cues. To suggests that ASDs have impairment in using contextual cues to moderate their assessment of emotional intensity, additional experiments for control group(typically developing children) will be conducted.

In developing the fifth core knowledge about social partners, race is an important factor biasing children to form social affiliations. The present study explored the development of the race-based social preferences in 3- to 8-year-old Taiwanese children. In Experiment 1, twenty-one 3-4 year-olds viewed three simultaneously presented video clips modeled by a Taiwanese (own race, high social status, in-group), a Southeast Asian (near race, low social status, out-group) and a Caucasian (other race, high social status, out-group) young female smiling at them. Children were instructed to give a toy to their most preferred and second preferred individuals, and children preferred the Taiwanese actress the most (50%). In Experiment 2, twenty-two 5-6 year-olds viewed the same videos and were instructed to choose their most preferred and second preferred persons as friends. They preferred the Taiwanese actress the most (69%). In Experiment 3, twenty-one 7-8 year-olds performed the same task as in Exp.2 and they preferred the Caucasian actress (57%) the most and the Southeast Asian actress the least. In sum, our findings suggest that a rudimentary race-based social preference (or prejudice) seems to emerge early in childhood. These results provide a cross-cultural exploration about the development of race-based social judgments.

Hills and Lewis (2006) reported that the own-race bias (ORB) in face recognition was reduced by cueing Caucasian participants to the lower region of African faces. However, recent empirical studies have failed to replicate this finding. This cross-cultural study investigated whether shifting initial attention to different facial regions affects own- and other-race recognition performance. In a classic yes-no recognition task, Malaysian-Chinese, Australian-Caucasian, and African participants were presented with a series of Chinese, Caucasian, and African faces that had facial regions (i.e., eyes, nose, mouth) cued with fixation crosses. Split analyses by race groups revealed that there was a pronounced ORB only in Caucasian participants. Most importantly, cueing Chinese participants to the nose region and Caucasian participants to the eyes and nose regions enhanced their subsequent recognition performance compared to cueing to the mouth region. These results suggest that immediate fixations directed to the upper facial regions, especially the nose region, somewhat encourage holistic face processing and/or decrease the engagement of featural processing, and may enhance the encoding of individuating diagnostic features for recognition.

Contribution of each facial part to attractiveness judgment of the whole face depends on exposure duration (Saegusa & Watanabe, 2016). In the current research, we examined how evaluator’s attention to detail would interact with the dynamic integration process of facial parts onto facial attractiveness judgment. Ninety-six participants evaluated the attractiveness of 98 whole facial photographs (58 female and 40 male models) and also isolated facial parts. The autism-spectrum quotient (AQ) was obtained from the participants. Multiple regression models explaining whole facial attractiveness with facial parts attractiveness were built for each participant. Then, we examined whether the sub-scores of AQ would show significant correlations with the contributions of facial parts attractiveness to whole face attractiveness. Our results demonstrated that, for participants who showed lower scores in the AQ sub-score of “focus to the detail”, the contributions of the eyes were lower. The finding indicates that the evaluator’s attention to detail interacts with the dynamic integration process of facial parts onto facial attractiveness judgment.

Recent studies revealed that persons with ASD showed reduced aftereffects in both facial identity and expression in adaptation paradigm. However, no previous study has examined whether these two aftereffects are independent or interdependent. The purpose of this study is to investigate ‘expression-contingent’ identity aftereffect (exp.1) and ‘identity-contingent’ expression aftereffect (exp.2) in ASD using an adaptation paradigm. The two experiments were done with 8 children with ASD. In experiment 1, two male faces with expression were used as adaptors, and 13 identity morphed images are used as probes. First, participants were shown the adaptor for 5000ms. Then, morphed images with either congruent or incongruent expression with the adaptor were presented for 500ms. Participants were asked to discriminate the ‘identity’ of the probe. In experiment 2, the procedure was identical to that of exp.1, except that ‘expression’ morphed images of either the congruent or incongruent identity with the adaptor were presented as probes, and participants were asked to discriminate the ‘expression’. Paired t-test was conducted to compare the accuracy between congruent and incongruent condition. There was no significant congruency effect in both exp. 1&2. These findings suggest that independent, not integrated face coding mechanisms of identity and expression in persons with ASD.

Contagious yawning—the urge to yawn when thinking about, listening to, or viewing yawning—is a well-documented phenomenon in humans and animals. While clinical studies have suggested the association between empathy and contagious yawning frequency, whether there is a perceptual component is not studied comprehensively yet. In this study, we examined influences from perceptual factors (i.e., individuals’ eye gaze pattern and perceptual detection sensitivity to yawning, happy, and angry faces) on 41 non-clinical adults. We induced contagious yawning with a 5-minute video and 20 yawning photo stimuli, and we measured participants’ eye gaze patterns, perceptual detection thresholds to human yawning and facial emotions (happy or angry), and autistic traits (with the Autism-Spectrum Quotient Questionnaire). We found two factors associated with yawning contagion: (1) those more sensitive to detect yawning, but not other emotional expressions, displayed more contagious yawning than those less sensitive to yawning expressions; (2) female participants exhibited significantly more contagious yawning than male participants. We did not find an association between autistic trait and contagious yawning after controlling for gender and yawning sensitivity. Our study offers a working hypothesis for future studies, in that perceptual encoding of yawning interacts with susceptibility to contagious yawning.

It has been reported that vertical disparities distributed in a certain spatial and temporal region are pooled to produce stereoscopic perception of the curvature, inclination, and slant of a surface. However, Duke & Howard (2005,2012) reported that vertical disparities in the same region are not pooled when they are separated in depth defined by horizontal disparity. In other words, vertical disparity is processed in each depth region specified by the horizontal disparity. We investigated the effect of a depth separation defined by relative motion, instead of horizontal disparity, on the vertical disparity pooling. The stimulus consisted of two sets of dots, and they were uniformly intermingled. They had opposite signs of vertical size disparity and opposite directions of motion. Their horizontal disparities relative to the screen were kept at zero. Observers responded perceived slants of one or two surfaces of the stimulus. As a result, when the relative speed of the two set of dots was high, two separated slants were perceived. When the relative speed was adequately low, single surface was perceived. The transition of slant perception seemed to be dependent on the relative speed itself, regardless of the magnitude of depth separation produced by the relative motion.

In this study, we measured how motion of the object is related to the visual attraction. We focused effects of the characteristics of the object in motion to the visual attraction in different condition in changing one of the parameters; number of elements in the object, element size, motion direction and a speed of the motion. We presented a pair of two different stimuli side by side and measured subject's eye movement in the observation to the stimuli. From the eye movement, we obtained the looking time to the left and right side stimuli. We simultaneously performed paired comparison between stimuli in the same presentation way and estimated the attraction intensity by z-scores of each stimulus calculated from selection rates. From these values, we firstly found that the highest attraction is obtained by the maximum number of elements (9 in 1 to 9), largest size (2.2 deg. in 0.44 to 2.2 deg.), upper-left direction in 8 directions, and fastest speed (7.3 deg./sec in 3.1 to 7.3 deg./s). We are now measuring the attraction in the new condition in which the size and number of elements are in trade-off relationship keeping the same area size.

Optic flow is generated whenever observers move relative to stationary objects in their environment. However, this optic flow is inherently ambiguous because the motion of environmental objects relative to a stationary observer can often produce similar patterns of visual motion. Many previous studies have shown that illusory experiences of self-motion (vection) can be generated when stationary observers view optic flow presented on a digital display. Most of these studies only considered random-dot displays where elements mimic the viewpoint-independent surface optics of diffuse reflectance. However, real-world surfaces also generate specular reflections that are viewpoint dependent. Very different visual motion velocities can be attributed to these specular and diffuse optic flows. In this study, we identified visual motion constraints in diffuse and specular optic flows that could be used to disambiguate object-motion from self-motion. We used the Oculus Rift to test whether vection varies differentially across conditions with and without specular reflections. The results support the view that the brain processes information about surface properties before visual motion is attributed to object-motion or self-motion.

Glass (1969) patterns are formed by pairs of random dots (dipoles) which could carry a powerful percept of global structure. The perceived motion direction of a dynamic Glass pattern is influenced by orientation of the dipoles (Krekelverg et al., 2003). Here, we investigated whether just updating random dots would form a specific global motion. Five types of moving stimuli, including linear, random, concentric, radial and updating, were shown through either a circle or square aperture. The task of the participants was to judge the perceive strength of concentric, linear, radial movement a 7-point Likert scale for each stimulus. In all viewing conditions, participants indeed showed a response bias toward concentric motion. The bias was particular strong in the circle aperture. Such effect of aperture shape was the most pronounced for the linear motion stimuli. These results show that just updating the position of dots in a random dot image is sufficient to generate a percept of concentric global motion. The effect of dipole orientation in the previous dynamic Glass pattern studies may be a secondary effect.

The fusiform face area (FFA) has often been speculated as a brain region that is specialized for face perception and recognition. While it is generally believed that the FFA responds selectively more to facial stimuli than other objects, the expertise hypothesis proposes that the FFA may participate in the processing of any object class that is trained to be processed at the subordinate or individual level. Next poster gives the FFA evidence of how two different training regimes (Gauthier et al., 1997, Vision Res., 37(12), pp. 1673-82; vs. Gauthier et al., 1998, ibid., 38/15, pp. 2401-28) yield different FFA responses. The multi-voxel pattern analysis is used to distinguish the patterns of FFA activity between Greebles and other stimuli (“Faces” and “Objects”), and we demonstrate that activity patterns of localized FFA perform better at distinction of “Faces vs. Greebles” in before-than after-training does, and in “Greebles vs. Objects” better in after- than before-training does. In both case, the Gautheir 97 paradigm has shown more prominent distinction results than the Gauthier 98 paradigm. In addition, searchlight information mapping is employed to identify other brain regions that can provide information concerning the neural representation of distinct object classes.

In one recent study, cortical thickness (CT) of car experts' Fusiform Face Area (aka FFA) were correlated with their face and object (car) performance [McGugin, et al. (2016) JoCN 28, pp. 282-294].To both extend this finding from car experts to experts of other domain, and also expand the CT and cortical volume,in study1 we reanalyzed our previously acquired birder MRI data (N=27 Caucasians), with both audiovisual and visual dprimes as their expertise measure.The results showed that significant correlations were found in both voxel density and CT between both audio and visual d', especially in bilateral fusiform gyrus,dorsal anterior and posterior cingulate gyrus, etc.After partialling out the age confound, controlling for the high correlation between expertise and age, these results still hold.In study2,we corroborated the similar results MRI data (N=20, with only visual dprimes) acquired in Taiwan.Lastly, the joint analyses combining both America and Taiwan data (N=47) showed that the left fusiform gyrus remained highly correlated,further strengthening the role of FG in expertise.Despite of slight disparities,the brain regions are overall highly similar across VBM- and CT-expertise correlations,not only extending the previous CT-expertise in car to bird experts,but also expanding the CT-expertise to VBM-expertise correlations,deepening the interconnection between experience and brain structure.

Is visual acuity for a two-letter stimulus the same or higher than for a one-letter stimulus? Previous experiments in central vision have shown that two characters if presented next to each other are more difficult to recognize than each character by itself. This is called crowding. We tested this finding with one-letter and two-letter stimuli presented in the periphery. Ten black-on-white letters ranging from 0.4 to 5.1 deg in height and presented at 10, 30, and 50 deg in the temporal visual field were used. Observation was with the right eye. The method of ascending limits was used. At 30 deg and 50 deg eccentricity, the threshold for recognition of two letters was markedly higher than for one letter, confirming the above hypothesis. At 10 deg results were obscured by a ceiling effect. The difference between 1 and 2 letters increased with retinal eccentricity. We attribute this difference to increased spatial interference. Results are consistent with Bouma’s (1972) critical distance rule, showing that two characters need to be spaced by a certain amount to prevent crowding. We kept the distance between pairs of letters constant, but increased their size relative to a single letter in order to make them recognizable.

The interocular grouping can occur when different parts of a visual stimulus are presented to each eye during binocular rivalry. The main purpose of this study is to test whether the dominant visual object from interocular grouping can induce the tilt illusion. The stimuli were three overlapped circles which had different sizes but same center. Each circle had square wave gratings which had different orientations. The two surround circles designed to be perceived as single grating which has a same orientation when the interocular grouping occurred. The integrated grating could induce repulsive or attractive effect of the tilt illusion. In the experiment 1a, the participants subjectively responded for the dominant duration and alternations of the interocular grouping. The orientations and sizes of two surround gratings varied. The interocular grouping were induced irrelevant of orientations and sizes. In the experiment 1b, the participants chose which direction the orientation of the grating in the center was tilted to. The result showed that the integrated grating induced the repulsive effect of the tilt illusion, but not attractive effect. These findings suggest that the attractive and repulsive effects of the tilt illusions during interocular grouping can be processed separately between conscious and visual processing levels.

Similarity and closure are two important cues for perceptual grouping. We investigated how these two cues interact with each other when they coexisted in an image. We used tripole Glass patterns which consisted of randomly distributed sets of three elements, including one seed and two context elements. Each local element had three bars arranged in U-shape. One context element was the same as the seed (similarity) while the other one was a 180o-rotated version of the seed (closure). The tripoles were arranged in a way that linking the seed with the one context would produce a clockwise (CW) spiral percept while the other one, counter-clockwise (CCW) spiral. The contrast of the context elements ranged from -20 to 0 dB while the seed contrast was fixed at -10 dB. Observers’ task was to indicate whether the test image was a CW or CCW spiral. We found that the observers were more likely (10%) to perceive a global pattern based on grouping by similarity than by closure. Such orientation similarity effect is consistent with a token matching theory of Glass pattern perception.

It is suggested that textures discrimination is mediated by a mechanism that is very sensitive to the lowest luminance in the texture. We investigated the role of spatial frequency on the function of such blackshot mechanisms in texture discrimination. We used two types of spatial frequency filtered random patterns. The first had 1/f-like spectrum with slope ranging from 0 to 1. The second was the bandpassed noise with 1 octave bandwidth and a peak frequency ranged from 2 to 32 cyc/deg. The white noise was constructed by a linear combination of the 2nd order Legendre polynomial and one of the other order polynomials. The observers were instructed to discriminate between textures based on the difference in the 2nd order polynomials. We estimated the contrast sensitivity profile by a maximum likelihood method. The contrast sensitivity profiles for all patterns peaked at the lowest contrast, consistent with the blackshot mechanisms. The blackshot sensitivity increased and then decreased with spectral slope, with peak at slope of 1. For bandpass patterns, the blackshot sensitivity decreased with the central frequency. Thus, the blackshot mechanisms are low passed and optimal for textures with a spectral slope similar to that of natural scenes

Measurements of threshold and reaction time are time consuming in practice and difficult for children with problems paying attention and difficulty controlling behavior. We measured the oblique effects using a rotatable disk for adjustment of line orientation. The observers were 15 young adults and 18 normal children. They matched the orientation of a line segment on the rotatable disk with that of a standard line segment. The orientation of the standard stimuli ranged from 0 to 157.5 deg (8 levels). Young adults and children respectively conducted 8 or 4 trials for each standard orientation, 64 or 32 trials in total. The mean errors were smaller for 0 and 90 deg than for 67.5 and 112.5 deg. SDs of the mean errors were also smaller for 0 and 90 deg. These results indicated that, compared with oblique line segments, observers’ judgements were generally precise concerning horizontal and vertical line segments. The mean errors and SDs were larger for children than for young adults. We consistently observed the oblique effects using the rotatable disk stimulus. It takes only about 10 min for children to complete 32 trials. Our apparatus could be useful for testing inattentive children.

Similarity and proximity are important grouping principles. A quantitative estimation of the relative strength of these principles is sparse in the literature. To address this issue, we manipulated these principles within a tripole Glass Pattern (tGPs), which consisted of random-distributed groups of dots. Each group contained one anchor dot and two context dots. An observer would perceive a clockwise (CW) spiral by grouping the anchor to one context dots and counterclockwise (CCW) spiral, the other. The luminance contrast of the context dots varied between -30 and 0 dB. The anchor-context dot distance varied between 2.5 and 20 min. Participants were to report whether they perceived CW or CCW in the tGP in each trial. In all conditions, the probability of seeing a CW spiral first increased then decreased with CW dot contrast increased. The peak of such inverted-U shape functions shifted rightward as CCW dot contrast increased. Manipulation of anchor-context distance showed a trade-off between similarity and proximity that peaked when the anchor-context distance was 5 min, which required about 6 dB contrast difference to canceled the proximity advantage. These results could be explained by a “divisive inhibition model” rather than the traditional energy or token-matching theory.

How do we effortlessly untangle complex mashes of image structure to visually infer the shape of 3D objects? We explored whether visual motion cues from different surface optics improve global shape perception. We created virtual objects varying in surface optics (matte, specular, refractive), relief (smooth, bumpy), and convexity along the viewing axis. Observers viewed stereoscopic simulations on the Oculus Rift of these objects oscillating horizontally for 5 s. The seated observers were asked to report whether the shape of the 3D object appeared elongated (like an Aussie football) or flat (like a pancake). We found that global surface convexity was perceptually overestimated for “flat” specular surfaces compared with diffusely-shaded surfaces (similar to Mooney and Anderson, 2014). However, we further found that “flat” refractive objects generated percepts of shape that were closer to veridical. We also found that elongated surfaces were underestimated in global convexity overall, but oscillating smooth elongated specular surfaces were perceived closer to veridical. The amplitude of oscillation improved shape discrimination for both smooth and bumpy surfaces. These results reveal a form of seesaw effect, whereby errors in perceived shape of objects differentially depends on motion cues from multiple forms of surface optics as curvature increases.