Contours and Contrast
Global and local factors in the enhanched checkered tilt illusion
Stefano Guidi, Oronzo Parlangeli and Sergioi Roncato
In the 'Enhanced checkered illusion' (Kitaoka, 2001, Nikkei Science, 31, 22-23) a horizontal line separating adjacent rows of checkerboard tiles appears tilted in presence of local inducers. This effect seems to derive from local contrast polarity relationships between the tiles and the inducers. Local factors, however, do not seem to be able to explain why the tilt seems greater when there are multiple rows of tiles and inducers (i.e. multiple tilted lines). In two experiments using 2AFC comparisons we studied the role of global factors in this illusion. Experiment 1 showed that the illusory tilt is increasingly stronger when the more lines are present. Results also showed that the tilt decreases as the horizontal distance between the inducers in a row increase. Experiment 2 showed that also the number of inducers in a row influences perceived tilt, which is stronger with more inducers. With multiple lines, however, increasing the distance between them seemed to increase the tilt. Overall, our experiments seem to show that purely local models cannot fully account for this illusion, and suggest that it might increase in strength with eccentricity.
Synergy of orientation, color and spatial frequency in shape detection
Guenter Meinhardt and Malte Persike
How features are combined is a basic question of perception research. Combinations of orientation (O) and spatial frequency (F) are special because they are effortlessly perceived in search displays, while combinations of color (C) and O require detailed scrutiny to be found among distracters (Sagi, D. (1988). Perception & Psychophysics, 43, 601-603). A possible reason for this is that O and SF are jointly coded on V1, while O and C are encoded at different layers of the hierarchy, suggesting a selective channeling - later integration processing scheme. We measured detection and identification of shapes defined by feature contrast in O, F, C and their combinations. In line with results from search experiments our data show that shapes defined by O+SF are much more salient than shapes defined by O+C. However, shapes defined by feature contrast in two spatially overlayed far apart spatial frequency bands (F1+F2) reached the same saliency levels as the O+SF shapes. Further, presenting the two frequency components dichoptically also yielded high shape saliencies at comparable levels. These results indicate that there is strong fusion across different V1 mechanisms. This implies that joint coding at the level of V1 is an unlikely candidate for explaining feature synergy in early shape perception.
Individual differences in the influence of contours on crowding
Steven Dakin and Calum Glen
We examined how crowding (the breakdown of object recognition in the periphery caused by interference from nearby visual structure) depends on the global arrangement of target and flanker elements. Specifically we probed orientation discrimination using a near-vertical target Gabor flanked by two vertical distracter Gabors (one above and one below the target). By applying variable (opposite-sign) horizontal offsets to the flanker positions we arranged elements so that on some trials they formed contours and on others they did not. While flankers generally elevated target orientation discrimination thresholds we observe maximal crowding not when flanker and targets were strictly co-aligned but when a small spatial offset was applied to flanker location, so that contours formed between flanker and target when the target orientation was cued. We also report that observers'orientation judgements are biased with target orientations appearing either attracted or repulsed by the global/contour orientation. A second experiment reveals that the sign of this effect is dependent both on the observer and on eccentricity. We describe a simple model of our results which computes perceived target orientation as a weighted/probabilistic average of target orientation and (a) flanker orientation and (b) contour orientation.
Visual perception of forms in 5-12 years old children measured by Form Test
Elena Bertossi, Sara Giovagnoli, Kerstin Hellgren, Roberto Bolzani and Mariagrazia Benassi
A new test of coherent form discrimination called Form Test is proposed in order to assess ventral stream functionality. The task consists of shape discrimination among 8 different forms. The form is obtained by a number of white luminance coherent dots presented on a circular black background. Stimuli are presented in 5 levels of spatial coherence by increasing noise dots. Noise dots are non aligned and non coherent luminance dots which disturb shape discrimination. A sample of 163 children (75 females and 88 males, age range: 5-12) is considered. Results show significant differences among coherence levels: the number of correct answers decreases with noise increment. Moreover, the age of children affect the scores: the performance of children in Form Test improves with age. The results confirm that Form Test can be an helpful instrument to measure the ventral stream functionality and shows its sensitivity to the age differences.
Effects of learning in amodal completion: an EEG study
Simon Jan Hazenberg, Marijtje Jongsma, Arno Koning and Rob Van Lier
We investigated the influence of learning on amodal completion, using EEG recordings. In a sequential matching task participants had to judge whether a test shape could be the partly occluded shape in a previously presented occlusion pattern. We distinguished between patterns that could evoke multiple perceptually plausible completions (ambiguous occlusion patterns) and patterns that evoke a single perceptually plausible completion (non- ambiguous occlusion patterns). Participants were tested before and after a learning task in which they had to learn and memorize the less plausible completions. In an earlier behavioural experiment we found learning effects only for ambiguous occlusion patterns. With regard to the current EEG experiment we analyzed ERPs elicited by the test shapes. Before the learning task, the test shapes elicited a P3 complex with comparable mean amplitudes and a maximum over Cz for both sets. For the anomalous shapes following non-ambiguous occlusion patterns no significant enhancement of P3 amplitudes was found due to learning, whereas for the ambiguous occlusion patterns a significant enhancement of P3 amplitude was found for the learned shapes only, which is in line with the behavioural data.
Poggendorff illusion with all-second-order contours
David Rose and Paola Bressan
The classic Poggendorff illusion is typically drawn with an outline or filled horizontal or vertical rectangle intersected by two diagonal lines. Misperception of alignment between the latter persists when the rectangle is defined by subjective contours interpolated between Kanizsa-style pacmen, or is a borderless area filled with parallel stripes. Recently (Rose and Bressan, 2011, Perception, 40 ECVP Supplement, 203) we reported that when the diagonals are defined by pacmen, the illusion reverses direction (for acute angles of intersection with the rectangle), for both outline and pacmen-defined rectangles. Here, we study the Poggendorff figure with the rectangle and its diagonals defined by stripe endings. To avoid the first-order luminance cues inherent in previous stimuli, we created a novel display with zero change in mean luminance across the entire display. The illusion was present with normal direction, indicating that the location cues presented by the first order stripe endings cannot be discounted and automatically participate in defining second-order contour locations at a stage preceding that at which the Poggendorff illusion itself is generated. In this respect, second-order contours defined by line endings are not equivalent to contours interpolated between pacmen.
Sensitivity to the Local Shape Information of Natural Images
Holly Gerhard and Matthias Bethge
We previously showed that the human visual system is exquisitely sensitive to local natural image regularities (Gerhard et al, 2011, 40 ECVP Abstract Supplement, 18). However, luminance histogram features were a very salient cue. Here we focus instead on sensitivity to the local shape regularities in natural images. Stimuli were textures made of image patches sampled either from natural images or from a natural image model, where observers had to discriminate the two kinds of textures in a forced choice task. We used a variety of natural image models that capture varying degrees of higher-order correlations. We removed salient luminance cues from the textures and measured discriminability of the models from natural images when shape was the only cue. Above chance performance with patches 4x4 pixels in size and larger indicated sensitivity to higher-order natural image correlations associated with shape. A surprising pattern of discriminability also emerged which indicated an advantage of the independent components analysis model in capturing salient shape content, even though it captures less of the overall higher-order correlations than other models tested. We also analyzed the contribution of different principle components to discrimination performance in order to develop a preliminary mechanistic explanation using responses of spatial frequency filter banks.
The role of motion information in contour integration: new insights
Rosilari Bellacosa Marotti and Clara Casco
The present study aimed at investigating the role of motion information in visual contour integration. We used a classical path detection paradigm to measure observers'efficiency at segmenting fragmented straight contours, composed of Gabor micropatterns, embedded in a field of distractor elements. Gabors were either static or moving, always drifting orthogonally to their orientation with constant speed. The elements forming the path varied in number from three to eleven and could be all oriented either parallel ('snakes') or perpendicular to the contour orientation ('ladders'). This way, they could be distinguished from random background on the basis of orientation alone, when static, or on both orientation and motion direction, either fixed or alternated. We compared detectability of paths when static, coherently moving or with their elements drifting in alternated directions. We found that motion facilitates detection of ladders (as previously shown) and also of snakes. The major finding was that alternating motion direction not only abolished motion facilitation but reduced detectability with respect to the static condition. These results suggest that motion has not a purely additive effect on contour detection but actively participates to the process of contour integration, probably throughout a cooperative motion mechanism.
Inferior superiority: shape discrimination is better in the lower compared to the upper visual field
Gunnar Schmidtmann, Graeme J. Kennedy, Harry S. Orbach and Gunter Loffler
Integration of information along the circumference of contours is an important feature of shape perception in central vision. To determine how shape discrimination and global integration depend on eccentricity thresholds for discriminating scaled radial frequency (RF) patterns with 3 or 5 lobes from a circle were determined at various eccentricities (0-10 deg) along the vertical, horizontal and main diagonal meridians. Global integration strength was measured by comparing performance when shape deformation was applied to various fractions of the contour. At 10 deg, sensitivity for all patterns was significantly higher in the lower than the corresponding upper visual field. For RF3, sensitivity in the lower, lower-right and lower-left positions were approximately the same and >50% better than for all other positions. For RF5, the lower periphery (lower, lower right and left) yielded sensitivities that are >2 times higher than the corresponding upper locations. The improvement in sensitivity when deformation was applied to the entire contour compared to a single cycle provides evidence of global pooling with similar strength at all locations. Discrimination of RF patterns is substantially better in the inferior than the superior visual field. The same global pooling strategy, affording the exquisitely high sensitivity to shape discrimination in central vision, also underlies peripheral shape processing.
Local density differences in perceptual grouping displays
Bart Machilsen, Johan Wagemans and Maarten Demeyer
Empirical studies on contour integration often focus on the collinearity of neighbouring elements as a grouping cue. During stimulus construction, differences in the relative spacing of contour and background elements are avoided as much as possible because they might serve as an alternate cue to the presence of a contour, invalidating any conclusions with regard to the collinearity (and co-circularity) of neighbouring elements. Avoiding local density differences is a challenging problem, however. We have implemented a number of element positioning procedures and local density metrics in GERT, the Grouping Elements Rendering Toolbox [www.gestaltrevision.be/gert; Demeyer and Machilsen, Behavior Research Methods, 2011]. Here, we investigated how sensitive human observers are to local density differences in arrays of radial Gabor patches. We present psychophysical data of a 2AFC experiment comparing two element placement methods and three local density metrics. We evaluate how sensitive each metric is to the physical cue, and how the metrics relate to human performance. The results can serve as a guidance for stimulus construction in future studies.
A convexity bias in peripheral contour integration
Maarten Demeyer, Bart Machilsen and Johan Wagemans
We present psychophysical evidence for a convexity bias in contour integration with respect to the fixation point. Participants were presented with a circular arc contour, embedded in a field of randomly positioned Gabor elements, and were instructed to respond as fast as possible whether the contour appeared to the left or to the right of the fixation point. Peripherally presented convex contours (i.e., contours that curve towards fixation) were detected faster than concave contours (i.e., curving away from fixation). We found this effect to be robust across multiple levels of eccentricity. These findings are at odds with the radial bias in visual perception, which predicts better detection for elements oriented parallel to a line through fixation.
Task differences in snake and ladder perception
Kathleen Vancleef and Johan Wagemans
In contour integration a relevant question is whether snakes and ladders are processed similarly. Higher detection thresholds for ladders indicate this is not the case. However, besides detection, discrimination tasks are widely used and a comparison between snakes and ladders in different tasks is lacking. In this study we investigate task dependency in snake and ladder processing by a direct comparison of tasks while using identical stimuli. Stimuli consisted of elements that are oriented collinear (snakes) or orthogonal (ladders) to the contour path and are surrounded by random oriented background elements. In two 2AFC-tasks, six experienced subjects either detected the contour when presented with a contour in one interval and a completely random stimulus in another, or they discriminated between contours of weaker and stronger curvature. Presentation time was varied in 9 steps between 8 and 492 ms. In general, the differences between snake and ladder processing were more prominent in the discrimination task than in a detection task. This pattern of results could indicate that ladders are processed by second-order mechanisms because more processing time is required especially when a finer shape judgment has to be made.
Psychophysical Indications of Recurrent Processing in Shape Perception
Jan Drewes, Galina Goren and James H. Elder
The human visual system must extract reliable shape information from cluttered visual scenes several times per second, yet the nature of the underlying computation remains poorly understood. Here we probe the temporal dynamics of this process to estimate time constants that might provide clues to the underlying neural circuit. In [Drewes, Goren & Elder, VSS 2012] we demonstrated a temporal window of facilitation in a repetitive-presentation contour grouping and shape discrimination paradigm with noise masking. Discrimination performance peaked at an ISI of 50msec, excluding iconic memory and probability summation as explanations for our findings. Here we report results of two new experiments that shed further light on this phenomenon. In Experiment 1, we replaced the first target shape presentation with a spatial cue. Discrimination performance did not exceed that of a single target presentation, demonstrating that facilitation is not the result of exogenous attentional cueing. In Experiment 2 we extended the range of ISIs further in time, finding evidence of temporal oscillation in behavioural performance at an average frequency of 21Hz, in agreement with our previous results. These findings suggest the existence of an important feedback loop in the neural circuit underlying the perceptual organization of contours into shape percepts.
Classification images for detection and discrimination of contours and shapes
Viljami Salmela and Miika Leppänen
We investigated the use of local features and global shape information in detection and discrimination of contour shapes. The stimuli were circular or square contours (diameter 1.34 deg; Gaussian profile, width 0.08 deg), or four evenly spaced contour fragments of the same shapes. The fragments located on the corners, on the sides or between the corners and sides. Classification images were measured for the detection of the stimuli and discrimination between the square and circular shapes. The stimuli were masked with Gaussian white noise (rms 0.27) and 1-interval yes/no and discrimination tasks were used. The adaptive Quest procedure kept the thresholds at 75%. The thresholds were lower for whole shapes than for contour fragments. The detection and discrimination thresholds did not depend on the position of fragments. The classification images, both in detection and discrimination tasks, contained significant information only at the location of contours. We did not find evidence of using global shape template with stimuli composed of contour fragments. The results suggest that shape perception does not depend on the location of the visual feature (corners vs. sides). Further, it seems that both detection and discrimination of shapes is determined by the local features of the stimulus.
Black or white? Effects of luminance polarity on target detection and discrimination
Marlis Gerdes and Cristina Meinecke
Textbooks often recommend black text on a white background (vs. white on black) to achieve maximum legibility for display users. This recommendation is supported by physiological studies showing that the visual system assigns more resources to processing stimulus elements with positive polarity (dark on light background) than negative polarity (light on dark background; [e.g. Ratliff, Borghuis, Kao, Sterling and Balasubramanian, 2010, Proceedings of the National Academy of Sciences, 107, 17368-17373]). Empirical evidence from applied and psychophysical studies, however, is mixed [e.g. Alexander, Xie and Derlacki, 1993, Vision Research, 33, 2491-2497; Buchner and Baumgartner, 2007, Ergonomics, 50, 1036-1063]. In a series of target detection and discrimination experiments, we tested the above recommendation. Two different measures of contrast were used to establish equal absolute values of contrast in positive vs. negative polarity conditions. Results varied depending on measure of contrast and task: in most cases we found no difference in detection/discrimination rates between the two polarity conditions. If there was an advantage for one condition, it was for the negative one. These results contradict textbook recommendations and predictions from physiological evidence. Alternative explanations are discussed.
Circle size judgment by psychophysical scaling
Marcelo Costa, Balász Vince Nagy and Adsson Magalhães
We used the magnitude estimation to obtain the apparent size of circles. Eighty-two subjects with normal or corrected-to-normal visual acuity (mean age= 22yrs; SD= 1.4) were tested. The procedure consisted of two gray circles luminance of 40 cd / m2, 10 degrees apart from each other. On the left side was the reference circle (visual angle of 1.1 cpd) in which was assigned an arbitrary value of 50. The subjects' task was to judge the size of the circles appearing in the right side of the monitor screen assigning the number proportional to the changed size, relative to the reference circle. Seven different sizes (0.6, 0.8, 1.0, 1.1, 1.3, 1.4, 1.5 cpd at 50 cm) between smaller and larger than the reference circle were presented. Our results have shown a high correlation (R= 0.9987) between the logs of the stimuli and the subject response. The exponent obtained was 0.71, which indicates that for the judged apparent size of the circle to be 2 times larger or smaller than the reference, we need a change of 3 times of the physical size. In conclusion, we found a no-linear judgment in the apparent size of gray circles in a visually normal population of young adults.
Developmental processes underlying orientation discrimination during childhood
Dave Ellemberg, Samuel St-Jacques, Mathilde St-Louis-Deschenes, Annie Baillargeon-Blais and Élise Labonté-Lemoyne
Orientation-selectivity is critical for higher-level visual processes from edge extraction to the perception of natural images. Only a few studies documented the development of orientation discrimination during infancy, and developmental changes during childhood are practically unexplored. The goal of the present study was to chart the development of orientation discrimination during childhood as a function of contrast, spatial frequency, and orientation noise to understand the changes in its underlying mechanisms during development. Stimuli were circular two-dimensional luminance Gaussian noise with a diameter of 5 degrees. The noise was filtered in the Fourier domain with an anisotropic filter. Orientation discrimination thresholds were determined by means of a one-interval forced choice paradigm combined with an adaptive staircase. Participants had to decide whether the grating was tilted to the left or right of vertical. Six-year-olds are about 3.5 times less sensitive than adults at low (1 cpd) and mid (4 cpd) range spatial frequencies and they are 4.6 times less sensitive than adults at higher (12 cpd) spatial frequencies. Eight-year-olds are still less sensitive than adults by a 2-fold, and that equally so for each spatial frequency. Orientation discrimination is adult-like at 10 years of age for all conditions tested, falling within the range of 0.5 to 1.0 degree. Orientation discrimination is contrast invariant in children when tested with low (1 cpd) to mid range (4 cpd) spatial frequencies that have a contrast equal to or greater than 25%. However, thresholds increase by a factor of two for a spatial frequency of 12 cpd presented at a contrast of 25%. The introduction of noise orientation increases discrimination thresholds for each age group, but significantly more so for the 6- and 8-year-old children. Orientation discrimination matures slowly during childhood and it is influenced by spatial frequency and contrast in the developing visual system but not in the mature visual system. Because internal noise explains much of the functional immaturity in children, the main source of developmental variation in orientation discrimination most likely occurs at the initial stages of feature coding.
Electrophysiological correlates of contrast perception
Piercarlo Mauri, Manuela Ruzzoli, Carlo Miniussi and Debora Brignani
Contrast information, a primary aspect of visual perception, has been usually bound to psychophysical investigation in humans. In the current work we recorded the electroencephalographic activity in healthy participants during a detection and discrimination task where a gabor was presented under six contrast levels. We analyzed how the variation of the contrast modulates the behavioural (i.e, accuracy and reaction times) and physiological responses, such as event related potentials (P100, N2pc) and evoked oscillatory activity. Results showed that both behavioural and electrophysiological indeces increased with a non-linear trend in relation to the contrast modulation, which increased exponentially through the conditions. While changes in the early visual P100 amplitude were more affected by exogenous stimulation, the late N2pc component varied consistently with the performance, appearing to be predictive of the behaviour. Moreover, the frequency analysis revealed a power increase of specific frequency bands such as theta (4-7Hz) and alpha (8-14Hz) in relation to the contrast variation. No modulation of the high frequencies was observed. On the whole, this study indicates that activity recorded over human visual cortex is related more to the subjects'percept than to the physically presented stimuli.
Orientation tuning of near and far surround modulation in V1 cells and in human perception
Nurminen Lauri, Shushruth Shushruth, Maryam Bijanzadeh, Simo Vanni and Alessandra Angelucci
We examined the orientation tuning of surround modulation in 5 human observers and in single cell (n=106) responses in V1 of anaesthetized macaques. For each cell a center grating patch was matched to the diameter of the cell's high contrast summation receptive field and optimized for other stimulus parameters; a surround annular grating was presented either near or far from the center grating, and its orientation was varied from optimal to orthogonal. Center and surround stimuli had 75% contrast. In humans, the center grating had a 2deg diameter, horizontal orientation and 1cpd spatial frequency, and was presented at 6deg eccentricity. We used a contrast matching task with center and surround grating contrasts of 20% and 40%, respectively. In both single V1 cells and human perception, the modulation arising from the near surround, i.e. within the reach of monosynaptic intra-areal horizontal connections, was sharply tuned for stimulus orientation. Instead, far surround modulation, i.e. arising beyond the extent of horizontal connections, was broadly orientation-tuned. Broader tuning of far surround modulation was due to non-optimal stimulus orientations exerting stronger suppression in the far than in the near surround. Our results suggest different orientation-specificities of the circuits underlying near and far surround modulation.
Searching in the dark - Saccades during scotopic and photopic visual search
Vivian C. Paulun, Alexander C. Schütz and Karl R. Gegenfurtner
When we search for visual targets in a cluttered background we systematically move our eyes around to bring different regions of the scene into foveal view [Najemnik and Geisler, 2005, Nature, 434, 387-391]. We explored how visual search behavior changes when the fovea is not functional, as is the case in scotopic vision. Five observers searched for Gabor patches (SD=0.5°, 2.5cpd, additionally 6cpd in photopic condition) embedded in a circular background (radius=8.5°) filled with 1/f noise under scotopic vs. photopic viewing conditions. Scotopic contrast thresholds were significantly higher overall with a functional scotoma in the fovea. Low spatial frequency targets were well visible in the periphery under both light levels. We found that in scotopic search individuals made less but longer lasting fixations and targets were located further in the periphery when detected. In both conditions fixations were dense in the upper center of the visual field. Their distribution was more widespread for higher spatial frequency targets in the photopic condition. This indicates that with adequate peripheral visibility observers prefer to make fewer saccades and take more time to fully integrate information over a larger window of attention.
Perceived speed and uncertainty: Can a Bayesian model of speed perception account for the effects of contrast?
Kevin Brooks and Kirsten Challinor
Bayesian models of motion perception propose that human perception of velocity is influenced both by signals from the stimulus and by a prior preference for stationarity. When velocity signals from the stimulus show little uncertainty, the prior has little effect, but as uncertainty is increased the prior exerts a greater influence, causing a more substantial reduction in the magnitude of perceived velocity. This model can successfully account for several illusions of perceived direction (Weiss et al., 2002, Nature Neuroscience, 5, 598-604) and has been extended to model the well known contrast-induced speed misperception - the phenomenon that low contrast stimuli appear slower than otherwise identical high contrast stimuli (Hürlimann et al., Vision Research, 42, 2253-2257). This model relies upon the assumption that relative to their high contrast equivalents, low contrast stimuli are accompanied by greater uncertainty. We tested this assumption by assessing the relationship between perceived speed and uncertainty over a range of spatial and temporal frequencies (2AFC speed discrimination: SFs 0.25-8c/deg; TFs 2-16Hz; speeds 0.25-64deg/s). While perceived speed was indeed reduced at low contrast, we find no evidence of the model's predicted negative relationship between PSEs and JNDs, calling into question its ability to account for contrast induced speed misperceptions.
Modelling grating and bandpass natural-scene contrast-discrimination dippers
Michelle To, Mazviita Chirimuuta, Edward Turnham and David J. Tolhurst
Chirimuuta, Jiwa and Tolhurst (2007, Perception: 36 Supplement, 157) presented a V1 multichannel model of contrast discrimination giving good fits for sinusoidal grating 'dippers' but poor ones for natural or 1/f images, and vice-versa. We now report contrast discrimination for 2.67 c/deg gratings, compared with photographic or 1/f images bandpassed (1 octave) to the vertical orientation and spatial frequency as the gratings. Gratings were presented as SL (Small Gabor test patch masked by Large grating), SS or LL; in-phase and out-of-phase annular grating masks on S grating test; and a crossed-orientation masking grating (SL). The 2 bandpassed images were configured SL and LL. A V1 model could be fit (3.6 dB2 error per point) to all grating and bandpassed image 'dipper' experiments simultaneously, provided the model's so-called 'non-specific' suppression had some orientation specificity. The model needed surround suppression to explain the different shaped 'dippers' for S versus L stimuli. Interestingly, surround suppression was still compatible with the finding that annular masking gratings had little effect on thresholds for surrounded S Gabor patches. The model fits were slightly improved by introducing extra features and their parameters to the model, such as adding a 'hard threshold' to the sigmoidal contrast-response transducer function.
Mirror symmetrical visual facilitation elicited by contingent auditory signals
In studying cross-modal interaction between auditory and visual processing, we found the enhanced contrast sensitivity with auditory primed contingent signals. Detection threshold of a Gabor visual target decreased when the 30-msec tone-burst preceded the target (Tanaka et. al 2009). Here I modulated stimulus duration (30, 50, 100, 300 ms), temporal frequency, and onset/offset patterns of auditory signals with various auditory-visual onset asynchronies. Threshold decreased with lesser amount (-0.1 log units) when the visual and auditory signals presented simultaneously. Threshold decreased more (-0.2 log units) when the visual target preceded the pip tone by 50ms, which was suppressed strongly and finally disappeared at 200 to 250 ms. Such oscillatory pattern was common across four observers tested with the variation of response frequencies around 9Hz. Facilitation disappeared using different temporal structures between vision and audition with different temporal duration and rising/falling time, suggesting that the cross modal temporal structure is critical to the effect. Since the oscillatory pattern is mirror symmetrical in time, I assume echoing signals traveling back and forth between visual and auditory systems, constructing periodical pattern of facilitation. This suggests the existence of double mirror neuron system between vision and audition mirroring signals each other in a different time.
Background influence on object detection in natural scenes
Justin Plantier, Stephane Buffat and Corinne Roumes
It is well established that the visual system analyses scenes through a series of spatial-frequency-tuned channels. Scene perception studies suggested that spatial frequencies of an image were analyzed from low to high spatial frequencies [Parker et al, 1992, Perception 21, 147 -160]. An alternative hypothesis is that starting with the retina, the earliest firing cells are those with the strongest inputs in term of luminance contrast [Delorme, et al., 1999, Perception, 28, 128-129]. From this hypothesis, the image is decomposed with a multiscale analysis with six bandpass filters, then a local spatial frequency is defined at each point of the image as the spatial frequency carrying the maximum luminance contrast. This analysis is used to characterize the object salience in function of background complexity. Images of natural scenes, and images reconstructed from the local spatial frequency map were briefly displayed (100 ms). In the experiment, the subjects had to detect the presence of a vehicle. The vehicles were positioned at different locations in the visual field. The object was always congruent with the scene context. The tasks performances (percentage of correct detection, response times) depend on the object salience and the background complexity. Results with images based on the local spatial frequency maps were similar to those obtained with initial images.
The effect of contrast on preference of print size
The purpose of this study was to investigate the effect of contrast on preference of print size. Four sighted persons read sentences on the CRT display and adjusted print size as they preferred. There were two display conditions (the POSITIVE condition: black text on white background and the NEGATIVE condition: white text on black background). In each condition, contrast was gradually decreased (95%, 50%, 25%, 10%). In addition, there were two vision conditions (the CLEAR vision condition: normal observation and the CLOUDY vision condition: observation with a diffusion filter). Results showed that, in both vision conditions, mean preferred print sizes were smaller in the NEGATIVE condition. The ratio of preferred print size in NEGATIVE/POSITIVE condition remained almost constant (0.79 to 0.75) at each contrast level in the CLEAR vision condition, while the ratio approximated to 1 as contrast decreased (0.52 at 95% contrast to 0.77 at 10% contrast) in the CLOUDY vision condition. These results suggested that, whether conscious or unconscious, we tend to prefer smaller print size in the NEGATIVE condition. In addition, it implicated that, for low vision persons with cloudy vision, the tendency was prominent at high contrast levels. [Supported by KAKENHI (#23730851)]
A reevaluation of area summation of contrast with compensation for retinal inhomogeneity
Alex S. Baldwin, Tim S. Meese and Daniel H. Baker
Increasing the area of a luminance-modulated sine wave grating decreases its contrast detection threshold. The process by which individual samples from discrete locations in the visual field are combined to achieve this is investigated here by analytic modeling. Several combinations and orders of transduction, template, and summation type were considered. Predictions from these models were compared to spatial summation results measured for two different stimulus types. The first was a set of circular sine-wave gratings (4 c/deg) of various diameters, including a subset of 'Swiss cheese' gratings that were modulated by a raised plaid to halve their total contrast over area [Meese & Summers, 2007, Proc. R. Soc. B, 274, 2891-2900]. The second set of stimuli were rectangular grating patches presented both in the fovea and in the periphery replicating Robson & Graham (1981, Vision Research, 21, 409-18). In other conditions, these stimuli were multiplied by an attenuation surface that compensated for the confounding loss of contrast sensitivity with retinal eccentricity. Our analyses reveal that the full wealth of our results can be described by a single model. This involves spatial filtering, square-law transduction and linear summation of signal and internal noise within a template matched to stimulus extent.
Spatial integration within and between first- and second-order stimuli
Robert Summers, Daniel Baker and Tim Meese
The detection of first-order (luminance-modulated - LM) and second-order (contrast-modulated - CM) stimuli is believed to involve separate mechanisms that interact weakly or are entirely independent; detection of an LM-only stimulus is barely improved by the addition of CM. However, little is known about the integration of stimuli comprising non-overlapping regions of LM and CM. Spatial summation of LM, CM and LM+CM targets was assessed using (i) full 1.25c/deg gratings of different sizes (1 - 16 cycles), (ii) fixed-diameter targets whose signal area was controlled by modulating a large (8 or 16 cycles) 'full'grating with a raised plaid pattern. The noise carrier (also present for LM stimuli) was bandpass-filtered white noise (8c/deg, ±0.5 octaves). We find that sensitivity improves with target size more rapidly for LM than for CM. When area was constant, comparing full and modulated stimuli yielded summation of ~5dB for both CM and LM. We also investigated cross-order summation, which was weak (~2dB) for full CM+LM (threshold adjusted) stimuli, but stronger (~3dB) when first and second order stimuli were interdigitated over area. This suggests a mechanism capable of integrating textures with attributes that vary over space, perhaps owing to changes in illumination or material properties.