Spatial Vision

Feature Integration in visual shape detection and discrimination

Malte Persike and Günter Meinhardt

Empirical evidence suggests that the visual system efficiently combines information from different feature cues to establish a stable perception of shape. The locations and procedures of mechanisms in the visual cortext responsible for feature integration are yet to be determined. We report results from two experiments employing a dichoptic presentation paradigm to show a) different shape defining parts or b) different shape defining features to each eye. Shapes consist of Gabor or line elements, having feature contrast against the background in orientation, spatial frequency, color, or pairwise combinations of these. Shape detectability and discriminatability from the single-cue conditions and double-cue conditions are measured in a 2AFC task. Performance in the double cue-conditions is then compared against various benchmarks derived from the single-cue conditions, such as probability summation and linear integration. Performance in the double-cue conditions from both experiments consistently exceeds the predictions based on the assumption of independent feature processing. Our results suggest that feature integration preceeds shape detection and is implemented early in the visual pathway. However, we find clear evidence that the integration of basic features like orientation and spatial frequency must occur in areas of the visual pathway where information from both eyes is already fused, thus rendering V1 an unlikely candidate for the site of joint feature processing.

Temporal dynamics of relative spatial frequency tuning of human perception

Tomoyuki Naito, Naofumi Suematsu, Eriko Matsumoto and Hiromichi Sato

In the present study, we investigated stimulus size dependency of spatial frequency tuning of human perception using psychophysical reverse-correlation method. All participants exhibited significantly stimulus-size-dependent absolute spatial frequency tunings (cycles/deg) such as increasing stimulus size decreased the peak spatial frequency. Then we calculated relative spatial frequency tuning (cycles) by multiplying absolute spatial frequency (cycles/degree) with stimulus diameter (degrees). We found that all participants exhibited almost perfect image-based spatial frequency tuning from response onset to response offset with fixed peak. Observed relative spatial frequency tuning was contrast invariant, although at low contrast (1%), optimal stimulus size was relatively larger than that at high contrast (90%), suggesting the most visible relative spatial frequency was contrast dependent. These results suggest that the relative spatial frequency tuning is a fundamental property of human perception.

Depth of Field and Visual Discomfort

Louise O'Hare, Tingting Zhang, Harold Nefs and Paul Hibbard

Cue conflict between depths specified by accommodation and vergence, which may arise under stereoscopic viewing, has been reported to cause discomfort [Shibata et al, 2011, Journal of Vision, 11(8), 1-29]. Depth-of-Focus (DOF) is a related depth cue [Vishwanath and Blaser, Journal of Vision, 10(10) 1-16] that occurs as a result of aperture (pupil) size and object distance to focal plane. Appropriate DOF increases performance on stereoscopically presented visual tasks [Hoffman and Banks, Journal of Vision, 10(5), 1-17]. In this experiment, we investigated whether incorrect DOF, as calculated based on the optics of the eye, causes discomfort. Stereoscopic photographs of objects with varying levels of DOF, too large, too shallow, or appropriate, were viewed whilst doing a visual dot counting task. Whereas increasing DOF blur increased headache and distortion ratings, surprisingly, more discomfort and eyestrain were reported with appropriate DOF than with inappropriate DOF. Reports of dry eyes were only affected by DOF in the presence of a simultaneous accommodation-vergence conflict. There was a peak in accuracy in a dot-counting task with appropriate DOF, in the absence of accommodation-vergence conflict. These results suggest that DOF is not useful alone in minimising discomfort, but can mediate effects of pre-existing cue-conflict.

Cityscape Impressions from Harbin Historical Postcards

Hiroshi Yamada, Nozomi Takahashi, Takahiro Nukui, Masashi Takahashi, Ryuta Suzuki, Hiroyuki Yoshida, Seiich Tani and Mitsuhiro Matsushige

We examined the cityscape impressions from historical postcards in terms of the semantic differential techniques. We used 695 Harbin postcards, which depicted three areas (Fujiadian, Pristan District, and Novi Gorod Street) of Harbin, China in the three different periods (before the Bolshevik revolution, before the Manchurian Incident, and the period of Manchukuo) from the Digital Archives of Asian Historical Material of Nihon University in Japan. The experiment was blocked by area and period of them. Each block showed ten cards which were randomly selected from the appropriate group of the cards by participant in random order. After observing them, each participant rated their impressions of the cityscape on each 5-point scale of thirty adjective pairs. A factor analysis of the ratings revealed five factors which were interpreted as activity, urbanity, solemnity, evaluation, and familiarity. We also conducted two factorial (three periods x three areas) MANOVA for five factor scores and subsequent tests. The results seem to correspond to some historical interpretations for three different periods of the three areas. We will discuss the possibility of psychological approach to some historiographical study based on these results.

Influences of temporally preceding contexts on boundary extension

Ayumi Egawa and Eiji Kimura

Boundary extension (BE) refers to a phenomenon in which a close-up photograph of a scene is remembered as containing surroundings that may have been present beyond the boundaries of the photograph. BE has been assumed to reflect an anticipatory processing; a limited view of a scene provided by a photograph is recognized within a likely spatial layout of the scene and the expected space is remembered as having seen. We reasoned that the anticipatory processing would be largely affected by preceding contexts, and tested this prediction. The stimulus composed of a successive sequence of three photographs zooming in or out on a scene and the last one was always a medium-angle photograph. BE was measured with a boundary adjustment task. The results showed that BE was much larger in the zooming-out condition, where successively appearing surroundings would have enhanced the expectation of the space beyond the boundaries and facilitated the anticipatory processing. Moreover, a smaller but significant BE was also found in the zooming-in condition, where the space beyond the boundaries was presented beforehand and thus the anticipatory processing would not have been activated. These findings suggest that the integration of successive views of a scene also contributes to BE.

Scene categorization at large visual eccentricities

Muriel Boucart, Miguel Thibaut, Christine Moroni, Michelle Greene and Sebastien Szaffarczyk

Studies on scenes perception have shown that the visual system is particularly sensitive to the global properties, or overall layout of a scene. Such global properties cannot be computed locally, but rather require relational analysis over multiple regions. To what extent does observers'perception of these global properties suffer in the far periphery? We examined how some global scene properties resist the decrease in spatial resolution as eccentricity increases from 10 to 70°. Pairs of photographs of scenes were simultaneously presented left and right of fixation for 80 ms on a panoramic screen (5 m diameter) covering the whole visual field. Central fixation was controlled. Observers were instructed to press the key corresponding to the spatial location left/right of a pre-defined target. We found that classification of global scene properties was accomplished with a performance highly above chance (around 70% correct) in the far periphery at 70°eccentricity. The perception of some global properties (e.g., naturalness) was more robust to the low resolution of peripheral vision than others (e.g., indoor/outdoor) that required a more local analysis. These results are consistent with the hypothesis that coarse features may capture the diagnostic image information needed for scene categorization.

Psychophysical Point: a disc becomes a point when Weber's law fails

Hiroshi Ono, Esther González and Linda Lillakas

In geometry, a point has a location but no area. It is an abstract concept and does not exist in the physical world. At ECVP (2008) we reported a series of experiments showing that the visual system treats a small disc (i.e., a dot) as a point. Here, we report the results of another experiment that shows the same. We placed the relative retinal image size cue for depth for dots and discs in conflict with the motion parallax cue to compare their accuracies with their jnd values. Pairs of dots and discs were presented on a computer screen; their movements were yoked to the observers'head movement (equivalent disparity of 0.08°). There was a 12% difference (above the jnd value for the dots) in the size of the dots (discs): 0.08° and 0.07° (1.5° and 1.32°). The results showed that the accuracy was higher for the dots (i.e., motion parallax was a more effective cue for the dots). We further confirmed the hypothesis that a dot provides information about its location but not about its area, because the discriminability of a dot as indicated by its Weber's fraction is low in comparison to that of a disc.

Sensitivity to numerosity is not a unique visual psychophysical predictor of mathematical ability

Marc S Tibber, Gemma Sl Manasseh, Richard C Clarke, Galina Gagin, Sonja N Swanbeck, Brian Butterworth, R Beau Lotto and Steven C Dakin

Individual differences in children's sensitivity to visual number (the ability to discriminate the numerosity of dot arrays) significantly predict mathematical performance. We examined the specificity of these findings in a broader population that included adults and asked whether other forms of visuospatial sensitivity similarly predict mathematical performance. Over 300 observers (6-73 years of age) performed a timed computer-based mathematics test followed by a batch of visual matching tasks, in which observers had to match two clusters of Gabor blobs with respect to their numerosity, density, size or orientation. Performance on all tasks undertaken (mathematical and visuospatial) improved with age, and there was a significant trend for better performance amongst male participants. When all visual thresholds, mathematics scores and self-reported education levels were subjected to a principal components analysis two significant factors emerged; one of these linked mathematics scores to general visuospatial sensitivity as measured by performance on all visual tasks tested. Further, although sensitivity to visual numerosity emerged as a significant predictor of mathematics scores once other parameters were controlled for, so did size and orientation acuity. These findings are consistent with visual numerosity tasks sharing resources with symbolic mathematics, but suggest that the critical factor is not numerosity itself.

Backward masking reveals different visual processing of schizophrenic and depressive patients

Maya Roinishvili, Eka Chkonia, Liza Reichard, Wenke Wurch, Hendrik Puhlmann, Cathleen Grimsen, Michael Herzog and Andreas Brand

Visual backward masking is a very sensitive tool for studying early visual processing deficits and a reliable endophenotype of schizophrenia. Mental diseases strongly overlap in many aspects, for example, in psychopathology, cognition, and genetics. Here, we show that strong masking deficits are found in patients with functional psychoses but not in non-psychotic patients, namely, depressive patients and abstinent alcoholics. We tested 28 schizophrenic, 22 schizoaffective, 20 bipolar patients, 26 major depressive patients, 23 abstinent alcoholics, and 24 healthy control subjects with various variants of the shine-through masking paradigm. Patients with schizophrenia, schizoaffective disorder, and bipolar disorder, show strongly prolonged SOAs compared to controls. Patients with unipolar major depression and abstinent alcoholics, however, perform like healthy controls. We suggest that patients with functional psychoses suffer from similar visual dysfunctions whereas visual processing of depressive patients seems to differ. The work was supported by the Volkswagen Foundation

Is texture segregation strictly feedforward?

G. Marina Veltkamp, Andreas Weber and Thomas Schmidt

Segregation of visual texture is a process driven by differences in the local spatial relations among elements similar in appearance. This process facilitates detection of region boundaries and is a key step leading to distinction of a figure from its background. It remains unclear whether this process can be solely ascribed to a feedforward processing stream or if recurrent or feedback information is needed. Using textures composed of random line arrays, and generated randomly for each trial, we employed a response priming paradigm to examine participants'ability to process textures within early vision. As predicted, local visual prime signals served to enhance/hinder visuomotor responses. In the absence of a local-defined prime the global features continued to influence speeded motor responses. Interestingly, a negative priming effect was observed in the control condition. In this case, the role of attention (task set) and perceptual learning is explored. Altogether these results reveal early processing of visual textures. By analyzing the time-course of in-vivo primed motor responses we make the case for a strictly feedfoward account of texture segregation.

Investigation of the Inattentional Blindness for Dynamic Events as a Result of Saccadic Suppression

Yuri Shelepin, Maria Kuvaldina, Alexei Harauzov, Olga Vakhrameeva, Sergei Pronin and Polina Yamschinina

The aim of our work - the investigation of the well-known Inattentional Blindness (IB) effect [Neisser, & Becklen, 1975, Cognitive Psychology, 7, 480-494; Simons & Chabris, 1999, Perception, 28,1059-1074]. Most demonstrations of the IB effect comprise the instruction to follow the target by the saccadic eye movements. This implies that IB can arise due to the saccadic suppression taking place in this task. In psychophysical experiments we analyzed eye movements during watching the classic and modified paradigm of dynamic IB tasks with different instructions. Instructions were supposed to evoke saccadic eye-movements with different parameters. We also varied the angular size of the testing images to analyze the importance of the perceptual area during eye fixations. The area of inattention depended on the instruction and the angular size of the test images. Obtained results are discussed in terms of different activity of Magno- and Parvo- system during saccades and fixations. We conclude that the IB effect in the classic dynamic IB paradigm could be explained not only by the mechanisms of attention but also by the well known effect named 'Saccadic Suppression' [Burr, Ross, Binda, & Morrone, 2010, Cell, 12 (14), 528-533].