Illusions

The Pot Lid illusion in parallel perspective

Stefano Mastandrea, Sherief Hammad and John M. Kennedy

One of us has frequently chosen a pot lid that was too big. Hypothesizing the error was a problem with foreshortening, we depicted a cylinder in parallel perspective, tilted 30°, with the top circle foreshortened to a 2D ellipse. Parallel projection provides foreshortening without convergence. Observers judged which of two 2D ellipses was identical to the one depicting the cylinder's top. One had the same dimensions as the ellipse depicting the cylinder top, 40 mm wide and 15 mm tall, and the other one was bigger, 41 x 18. The observers chose the bigger ellipse. An ellipse depicting a cylinder's top may look bigger because of crosstalk from perceptual information for a circle in perspective. This information for a tilted 3D surface 'biases' perception -- merely biasing the 2D ellipse towards the circle. The 2D ellipse's apparent eccentricity is simply lessened -- it does not go all the way to a perfect circle. With respect to our original errors with real pot lids, if depicting ellipses can be biased towards circles then surely really circular pot lids can be similarly biased towards ellipses, when the lids are viewed highly tilted.

Modification of each color component and shaking to enhance optimized Fraser-Wilcox Illusions

Kazuhisa Yanaka, Teluhiko Hilano and Akiyoshi Kitaoka

We describe a new method to enhance Kitaoka's optimized Fraser-Wilcox optical illusions other than Type V by using only red and blue color components and shaking the image. The Fraser-Wilcox illusion is very interesting because it causes illusory motion although it is actually perfectly stationary. The amount of motion is considerably increased when a soft or a hard copy of an illusory image classified as Type V by Kitaoka is shaken at the frequency of several Hz (Yanaka et al., ECVP 2011). However, as optical illusions other than this one are cannot be strengthened by shaking, we have introduced a new technique. When the original illusionary image was already colored like Kitaoka's 'Rotating snakes', the green component was completely removed, and the gray levels of the red and blue component were corrected respectively. When the original illusory image is monochrome, the image was colored by using tone curves of red and blue components although the output of the green component is always set to zero, depending on the level of grayness of the original image. When the modified images were shaken, strong illusionary motion was observed. This technique can be applied to not only 'Type V' but also almost all types of Kitaoka's optimized Fraser-Wilcox optical illusions.

Okuma illusion: a new illusion of leaning

Takashi Yasuda, Takashi Ueda and Kenpei Shiina

A new illusion about tilt perception of objects is reported. Physically, the objects in a rotated picture should tilt the same degree as the picture itself. Our finding is, however, that the tilts of objects in a rotated photograph are perceived differently. The picture used in this illusion contains a statue and a tower. When the picture is tilted, we perceive that the tower follows the tilting of the picture, while the statue doesn't. We tested several objects and found that this phenomenon was salient when an element in the picture had non-geometric shape. The same inclination effects were observed when the statue was replaced by another statue or an animal, whereas the effect was diminished when the statue was replaced by a building. This illusion is possibly caused by the difficulty of judging the centerline of the object. If the centerline of an object is easy to estimate, as in the case of a tower or a building, our perception of tilt will be correct. If an object has a non-geometric shape, the detection of centerline will be difficult, making our perception of tilt harder.

The reverse Café wall illusion

Midori Takashima, Ken Shiina, Teruo Fujii and Kouki Shinohara

When the mortar lines in Café Wall Illusion were wider, the perceived direction of tilt of the mortar lines is reversed (Earle and Maskell, 1993, Perception, 22, 383-390). However, they did not mention the parameter in detail. We studied how the width of mortar lines affects the direction of tilt in Café Wall Illusion, Munsterberg Illusion, and 'reverse Café Wall illusion' (Kitaoka, Pinna and Brelstaff, 2004, Perception, 33, 11-20). As our results, when gray mortar line in Café Wall Illusion was a little wider we clearly perceived the reverse illusion, but when the mortar line was further wider the illusion decreased. When black mortar line in Munsterberg illusion was wider, the reverse illusion was apparent but weak. When the mortar line in 'reverse Café Wall illusion'was wider, perceptual direction of line tilt could be further reversed, that is in the orthodox direction. Some participants did see reversal of the reverse Café wall illusion but some did not. Our results cast the caution for the theories and models to explain these illusions.

Anomalous Motion Illusions and Retinal Visual Reset

Masanori Idesawa

Anomalous motion illusions observed in still figures (Fraser and Wilcox, 1979 Nature 281 565-566) and in swinging figures (Ouchi illusion family) are the fascinating topics in vision study. In the former study (Idesawa, 2010 Optical Review, 17, 557-561, Idesawa, 2010, Perception 40, ECVP Abstract Supplement 172), the author had proposed the model for explaining them by postulating the filtering process (sharpening with lateral inhibition and blurring in image slip) and the visual reset (periodical zero level renewing); then, simulated them successfully. The filtering process can be expected easily from the features of integration, delay and lateral inhibition in the retinal cells. For the visual reset, the horizontal cells, in which short term zero levels in certain areas may be generated by spatiotemporal integration and fed back with delay, may play important role. The author has examined the rate of periodical reset by detecting the velocity giving stationary perception in the moving grating patterns; then found that it was varied with the environmental illumination level (for instance, 25lx:~20Hz, 500lx:~35Hz, 5000lx:~40Hz ....). It is inferred that the feedback delay has some relations with the period of visual reset and the synchronization may take place often with micro-saccade, blinking and/or cortical level reset.

The Fraser-Wilcox illusion and its extension

Akiyoshi Kitaoka

The Fraser-Wilcox illusion [Fraser and Wilcox, 1979, Nature, 281, 565-566] is one of anomalous motion illusions displayed in a stationary image. Here I review studies of the Fraser-Wilcox illusion and its extension including 'Rotating snakes' (http://www.ritsumei.ac.jp/~akitaoka/index-e.html). Perceptual dimorphism featuring the Fraser-Wilcox illusion is explained by the interaction between two coexsisting illusions. A new effect, 'darkening disambiguation' of the Fraser-Wilcox illusion, is proposed. Moreover, the 'optimized' Fraser-Wilcox illusions and the role of color in these illusions [Kitaoka, 2006, Paper presented for the 9th L'ORÉAL Art and Science of Color Prize] are discussed.

The relative motion aftereffect in the Enigma illusion

Erika Tomimatsu and Hiroyuki Ito

Enigma is an artwork created by Isia Leviant, in which illusory motion is seen in rings placed on a radial pattern consisting of spokes. We tested whether a relative motion aftereffect could be reflected in the illusory stream on the ring. Participants viewed the center of a physically rotating Enigma stimulus as adaptation. We found that when spoke rotation was stopped, an illusory stream on the ring was seen mainly in the same direction as that of the preceding physical spoke rotation. In a second experiment, we replaced the gray ring with a ring filled by random-dots. To investigate the effect of the relative motion adaptation in the Enigma illusion, we rotated the random-dots and the spokes independently in the adaptation period. When a static Enigma image was presented after viewing the ring and spokes rotating in opposite directions, an illusory stream on the ring was vividly seen in the same direction as that of the preceding spoke rotation. The results suggest that relative motion detectors with receptive fields corresponding to the ring area produce the illusory motion stream in the Enigma illusion.

The Spider illusion

János Geier and Mariann Hudak

We show a dynamic illusion. A section of an Ehrenstein pattern (Fig.1 in Pinna B, Spillmann L, Ehrenstein W H, 2002, 'Scintillating lustre and brightness induced by radial lines' Perception 31(1) 5 - 16) is moved up and down taking 1 cm long steps in a 5 cm wide, 20 cm long vertical path. This way, an illusion is obtained, which is similar to a spider clambering up along the two vertical edges of the path, while its legs are moving downwards. It can also be described as if two half cogwheels were rotating in opposite directions along the edges. When the figure is moved downwards, the 'spider legs'or 'cogs'are seen as if they were moving upwards. The illusion is even more effective when Stubb's figure is used (Fig. 6 in Gori S, Stubbs D A, 2006, 'A new set of illusions-the Dynamic Luminance-Gradient Illusion and the Breathing Light Illusion' Perception 35(11) 1573 - 1577). The possible explanation is identical with that of the phi-effect: if an endpoint of a horizontal 'spider leg'abruptly jumps to the previous location of the corresponding endpoint of an oblique one, then it is perceived that the horizontal line has become oblique.

Illusory motion and motion capture in terms of oblique components.

Makoto Ichikawa and Yuko Masakura

When viewing the concentric circles, which consist of oblique components, the observers see an illusory rotation of the circles by changing the viewing distance (Pinna and Brelstaff, 2000, Vision Research, 40, 2091-2096). If several dots were superimposed on the concentric circles, they will see the illusory rotation not only for the circles, but also for the dots (Ichikawa et al, 2006, Perception, 35, 933-946). This illusory rotation of the dots, which have no means for generating illusory motion themselves, is based on 'motion capture'. In this study, we examined how the amount of dots (ranging from 0 to 32) and proximity among them (ranging from 1mm to 110mm on a paper) affect the illusory rotation for the circles (approximately 150mm in diameter) and motion capture for the dots. Although the illusory rotation for the circles declined with the increment of dots, that for the dots increased with the increment of dots. The proximity among the dots has no consistent effect for motion capture, while it enhanced the illusory rotation for the circles. These results suggest that motion capture is caused by the leakage of motion signal from the oblique components, and accumulation of the motion signal at each superimposed element.

The influence of motion integration on shifts in perceived position

Peter J. Kohler, Caeli E. P. Cavanagh and Peter U. Tse

The perceived position of a briefly flashed stimulus can be shifted in the direction of nearby motion [Whitney & Cavanagh, 2000, Nature Neuroscience 3(9) 954-959]. We used a horizontally translating diamond stimulus [Lorenceau and Shiffrar, 1992, Vision Research 32(2), 263-273] to examine how different interpretations of ambiguous local motion affect the position shift. Two dots were flashed as motion direction reversed, such that the dots appeared shifted in opposite directions. Subjects viewed the stimuli while maintaining central fixation, and adjusted comparison stimuli to report perceived dot positions. When the diamond was presented behind occluders producing a vertical global motion percept, the dot position shift was primarily vertical. When the diamond was presented without occluders, the global motion was horizontal, but surprisingly, the dots were shifted both horizontally and vertically. In both cases the local diamond motion at the dot positions was identical. Position shifts can also occur in the absence of physical motion signals [Watanabe, Nijhawan and Shimojo, 2002, Vision Research (42), 2645-2650; Shim and Cavanagh, 2004, Vision Research (44), 2393-2401]. Here physical motion is present but ambiguous, and we find that position shifts are influenced, but not completely determined, by the integration of physical local motion into global percepts.

On visible persistence: inter and cross modal measurements in transient random dot stimuli

Kathrin Thaler and Maximilian Bruchmann

Visible persistence refers to the 'extension of the visual stimulation beyond the termination of the stimulus itself' [Bowen, Pola, & Matin, 1974, Vision Research, 14 (4), 295-303]. The aim of this study was to investigate minimal persistence duration and variables that might affect it using stimuli with infinitesimal duration. We used random dot kinematograms (RDKs) where on the transition from one RDK to a second one all pixels in an annulus-shaped area changed luminance polarity, creating physically a single second order luminance transient. Phenomenologically, this transient signal typically produces the percept of an annulus with an abrupt onset and a gradual offset. We varied the radius and thickness of the annulus and measured its visible persistence by letting subjects match the onset of an auditory or visual reference stimulus to the onset and offset of the illusory annulus. The results showed that auditory reference stimuli yield longer persistence estimates than visual ones. Furthermore, persistence increased with stimulus size. We discuss whether this increase is better explained by a simple function of stimulus energy or by a filling-in mechanism.

On the problem of shape formation: New effects related to Oppel-Kundt and Helmholtz's square illusions

Maria Tanca and Baingio Pinna

According to the Oppel-Kundt illusion, an empty (unfilled) space looks wider than a space filled by some objects (Oppel, 1854-1855). This illusion is related to Helmholtz's square, where a square appears wider when it is filled with vertical lines and higher when filled with horizontal lines (Helmholtz, 1866). Recently, Pinna (2010, 2011) demonstrated the rectangle illusion, according to which, given a large square shape made up of small squares, the grouping of the small squares on the basis of the similarity principle influences the shape perception of both small and large squares. The direction imparted by grouping is the attribute that influences the shape by polarizing it in the same direction both globally and locally. The rectangle illusion is opposite to what expected on the basis of Oppel-Kundt and Helmholtz's square illusions. Aim of this work is to solve the antinomy between the two sets of illusions and to demonstrate a common explanation based on the interaction between different sources of directional organization. This was accomplished by introducing some new phenomena and through phenomenological experiments proving the role played by the directional shape organization in shape formation. Supported by Finanziamento della Regione Autonoma della Sardegna, ai sensi della L.R. 7 agosto 2007, n. 7, Fondo d'Ateneo (ex 60%)

Changing shape by changing accents in space and time: Visual and musical shape illusions

Luca Sirigu and Baingio Pinna

Pinna & Sirigu (2011) demonstrated a new principle shape formation called the accentuation principle, stating that, all else being equal, the elements tend to group in the same oriented direction of the element discontinuity placed within a whole set of continuous/homogeneous components. The discontinuous element is like an accent, i.e., a visual emphasis within a whole. It was shown that the accentuation principle can influence grouping and shape formation not only vision but also in music perception. Aim of this work is to study the role of simultaneous multiple sources of accentuation (synergistic and antagonistic), including the rhythm, in forming visual or musical shapes. Some new illusions, where the same stimulus pattern can be perceived like different possible and as alternative visual or musical shapes, are studied through phenomenological and psychophysical experiments. The results demonstrated that the accentuation principle strongly influences not only the shape but also the grouping and the figure-ground segregation in both visual and musical domains. Our outcomes weaken Ehrenfels's general law according to which we can recognize a melody as one and the same even though it has been transposed into a different key or has been played on a different instrument or at a different speed. Supported by Finanziamento della Regione Autonoma della Sardegna, ai sensi della L.R. 7 agosto 2007, n. 7, Fondo d'Ateneo (ex 60%)

The elaboration of "double overturn" method for the research of visual image formation

Olga Arbekova and Alexey Gusev

The new method for research of visual image formation was developed. The method allows separate three components of perception: sensory 'fabric', meaning and personal sense [Leontev, 1975, Activity, Consciousness, and Personality, Moscow, Politizdat]. The participants looked at the doll`s room through the invertoscope. This toy room was really overturned and all interior items were glued. So we named our method «double overturn». The participants were asked to describe the visual scene and to do actions in the space of toy room (to put the spoon into the cup). The experimenter could manipulate with things in the room too (pour out the sugar from the teapot to ceiling of overturned room). It gets the opportunity to oppose the sensory 'fabric' and the meaning of perceived scene against objective reality whereas in famous experiments with the invertoscope the sensory 'fabric' and the meaning were opposed. The results with 59 participants showed: 1) the observers didn`t notice the scene was really overturned; 2) to encourage them to notice the overturn, the sensory 'fabric' and meaning have to be collided; 3) for this purpose, both observer`s and experimentator`s actions were an equal success; 4) a number of illusions were observed in the both cases.

The sensorimotor measurements of perceptual illusions

Valeriia Karpinskaia and Vsevolod Lyakhovetskii

The role of interhemispheric asymmetry at the sensorimotor measurements of Muller-Lyer illusion was investigated. We used the illusion of perception as a phenomenon of consciousness to find out how the error's value depends on the left and right hemisphere contribution to the control in decision-making. We compared the precision of the left and right hand's movements on the touch screen during the measurement of the segments length. It was found that the left and right hand controlled by two hemispheres by different ways had made different errors during these measurements. The constant bias of the right hand's movements is larger than for the left one. The conclusion is that the hemisphere mostly involved in movement control determines the dominant system of representation (metric and categorical) and the control's type in decision-making (automatic or conscious).

The early time course of the Delboeuf illusion

Filipp Schmidt, Thomas Schmidt and Andreas Weber

In the Delboeuf illusion the perceived size of a circle is modulated by the size of a surrounding annulus. Because only little is known about the early time course of this effect, we studied the illusion within a response priming experiment. In each trial, we simultaneously presented a small and a large peripheral target circle on the left and right side of fixation. Participants responded to the position of the one or the other by pressing a left or right button. Shortly before the targets we presented small, large, or medium-sized central primes that were consequently consistent, inconsistent, or neutral with respect to the required response. In the critical condition, neutral primes were surrounded by annuli rendering them smaller or larger. Importantly, this manipulation produced response priming effects as would be expected if the illusion was effective also in fast visuomotor processing. However, the magnitude of these effects strongly depended on response speed, with larger priming effects in slower responses. We discuss our findings within the framework of a dichotomy between a temporally early, coarse and global visual processing style versus a late, fine and local one. Only the latter would be able to integrate illusory percepts.

The parallelogram: A perceptual enigma.

Fred Halper

Veridical perception of planar nonrectangular parallelograms is difficult, with new paradoxes of organization recently observed [Tyler, 2011, i-Perception, 2, 137-141]. A Sunoco gas station canopy of such shape in New Jersey, 40°50'18.28'N 74° 09'23.76'W [©Google (2010) Google Earth (v.6.0)] is a real-world example. Four subjects shown 10 surrounding views, all reported with strong confidence, the canopy as rectangular. Presenting single images, the canopy is accurately seen by only 7% of 116 subjects, making a choice from 4 alternative shapes, as viewed from above. Eighty percent of subjects pick a rectangle or square, with the remainder opting for quadrilaterals. Although the canopy is level with the ground, 20% of subjects report it to be tilted. This is less robust than the illusory tilt of attached Manhattan balconies of the same shape [Halper, 1997, Perception, 26, 1321-1322], where the vertical surface of the building provides relative information about right-angularity. Another group of subjects viewed photos of a Sunoco canopy whose shape is square. Although veridicality tripled, rectangles were chosen 58% of the time, with the total of right-angled shapes perceived being the same (81%) as the parallelogram. A conclusion is that gravity, in nature and architecture, pushes the visual system to organize four sided figures as right-angled.

From physical to perceptual stimulus representations: Exploiting the Ebbinghaus illusion in a response priming paradigm

Andreas Weber, Lisa Noé, Lukas Hoffmann, Thomas Schmidt and Filipp Schmidt

Stimulus representations change over time in visual processing from early stimulus-governed representations to later perceptual interpretations. The study of the time-course of visual illusions is an insightful approach to follow this continuum within visual perception. In the Ebbinghaus illusion the perceptual size of an object is increased/ decreased if it is surrounded by smaller/ bigger context elements. In a response priming paradigm, participants reacted to the position of the smaller/ bigger of two simultaneously presented target circles. Target stimuli were preceded by prime circles whose perceptual sizes were modulated by different versions of the Ebbinghaus illusion. By rendering the primes smaller or bigger, the illusion led to response-consistent or inconsistent configurations of primes and targets. Regression analyses revealed a continuum between two qualitatively different result patterns: Slow participants/ trials showed priming effects in the direction of the illusory percept while fast participants/ trials showed opposing priming effects. Thus, within our measurement we were able to completely record the continuous transition from physical to perceptual stimulus representations. Inferences about the respective properties are discussed.

Perceived display speed helps account for the 'jitter advantage' in vection

Deborah Apthorp and Stephen Palmisano

Illusions of self-motion in depth ('vection') are strongly enhanced when coherent viewpoint oscillation or jitter is added to the inducing optic flow displays (Palmisano et al., 2010, Perception, 29, 57-67). The underlying cause of this 'jitter advantage' is still unknown. Here we investigate the possibility that perceived speed plays a role, since other manipulations that increase perceived speed (adding stereo, using contracting rather than expanding flow) also increase vection in depth, and reducing perceived speed reduces vection. First, in a 2AFC procedure, we measured PSEs for smooth and vertically oscillating motion-in-depth displays; oscillating displays were uniformly perceived as faster. Then we used throttle and verbal measures to measure strength, duration and onset latency of vection in oscillating and smooth linear vection displays matched for real and perceived speed. The oscillation advantage was eliminated when displays were matched for perceived speed. Adding motion blur increased vection while leaving retinal speed constant. We tested whether observers were misattributing vertical motion to MID in a further 2AFC estimation of oscillation magnitude, but this was not the case. Testing with random viewpoint jitter, we found that jittering displays produced greater vection than speed-matched displays, suggesting there may be separate mechanisms driving jitter and oscillation effects.

Texture-region analysis at the basis of the illusory enlargement of a filled area

Enrico Giora and Simone Gori

Similar to what happens in the Oppel-Kundt phenomenon, an area filled by clearly distinguishable subparts appears larger than an empty space of identical physical size. The perceptual overestimation depends on textural properties and results to increase with spatial frequency and decrease with the microelements'number (Giora and Gori, 2010, Vision Research, 50(23), 2466-2475). Two alternative approaches account for texture processing: edge- and region-based models. In the present research we investigate whether the phenomenon of area enlargement is based either on an edge-based analysis, focused on the boundaries'information, or on a holistic processing of all the texture-region. We tested the perceived size of figures filled by textural micropatterns either only at the boundaries or in the entire texture. Textured stimuli were compared to uniform figures, whose size observers had to adjust. Results showed that the perceived size increased proportionally with the amount of textured area and confirmed the crucial role of spatial frequency and microelements'number. We concluded that perceiving the size of visual objects involves a holistic analysis of textures, sensitive to the statistical properties that are processed throughout the entire texture-region.