CONTRIBUTIONS TO CONTRAST AND MOTION VISUAL ILLUSIONS
JACQUES NINIO
(included in the web site http://www.lps.ens.fr/~ninio
TOPICS DISCUSSED HERE:
Extinction effects
Flashing lines
Orientation-dependent contrast
Ouchi variants
Hulla-hoop illusion
Gliding circles
Colour-emitting wheel
SEE IN OTHER SECTIONS
- Geometrical illusions (Contributions to geometrical illusions)
- Camouflaging textures
- Depth effects with a single eye
++++++++++++++++++++++++++++++++++++++++++++++++
OVERVIEW
My main interest, in the field of visual illusions has always been, starting in 1975, understanding the geometrical principles underlying geometrical visual illusions. Nevertheless, I took part in the design of several new contrast or motion visual effects, the best known being the "extinction effect", co-authored with Kent Stevens. The discovery of the new effects was produced by the conjunction of several factors:
============================
EXTINCTION EFFECT
Figure 1a (pdf): Extinction effect, black background
Figure 1b (pdf): Extinction effect, white background
Very few people in the field of vision research have a serious background in geometry. Jan Koenderink, Kent Stevens Johan Wagemans or Jacques Droulez are among the rare exceptions. I met Kent Stevens at a meeting on stereo vision organized by Ian Howard in Toronto in 1993. I knew some of his classical papers on how the brain interprets 3D from line drawings (for instance [3]) and we had a nice discussion at the welcome cocktail. In 1997, I was in the USA as a tourist, and I seized this occasion to pay a visit to his lab at Eugene, in Oregon. On the same day, I paid a visit to Frank Stahl, a molecular geneticist at the same university , both of us being interested in unorthodox mutation mechanisms, and the so-called problem of "adaptive mutations" — See Section on molecular evolution). I suggested to Kent that it would be nice if he came to Paris for a few months or a sabbatical period in my laboratory at Ecole Normale Supérieure. He seemed pleased at the prospect. Ultimately the trip materialized. Kent spent a month in Paris (June 14th - July 13th, 1999). At that time, he had nearly abandoned vision, and was working on modelizing the posture and the gait of dinosaurs.
Since Kent had worked in the past on the link between receptive field modelling and contrast illusions of the Munsterberg family, in which there is a strong geometrical component ([4]), I proposed to him to investigate, in the same spirit the link between possible neurophysiological architectures and another contrast illusion — the Hermann grid illusion.
At the time of our study, the state of the subject was the following: The Hermann grid was known in its standard form, with squares separated with horizontal and vertical alleys, and a number of fine observations had been made on the subjective grey-level distributions (review in [5]); Bergen had described a very spectacular phenomenon at an ARVO meeting in 1985: when a Hermann grid is made fuzzy by the removal of the high frequency components in the image, the crossings of the alleys start to scintillate [6]. The phenomenon was easily demonstrated on a computer screen, but Bergen did not produce a convincing paper version of his illusion. Then Schrauf and co-workers [7] produced a modified Hermann grid with black squares separated by grey alleys, having white disks at the intersections of the alleys. Scintillation could be seen inside the disks. There were mathematical models of the Hermann grid [8] but these were based on the square lattice geometry of the grid, and since this configuration had not been shown to be in any way necessary for the effect, these models counted for nothing.
Both Kent and I were aware of the existence of Schrauf’s scintillating grid, but we were unaware of Bergen’s contribution. On the first or the second day of his visit, Kent produced a fuzzy Hermann grid, and rediscovered Bergen’s effect. We were both very excited, but ultimately, reading carefully Schrauf’s paper, I found there the reference to Bergen.
In any event, we had at least two phenomena to consider: the pure Hermann grid effect, and the scintillation effect, and did not know whether they were two aspects of a same phenomenon or not. So I proposed to study systematic geometric variations of the Hermann and Schrauf grids in parallel, trying to determine whether or not there were common geometrical requirements for the two phenomena (for instance, could it be the case that if one replaced the square lattice by a triangular one, one of the two effects would persist, while the other effect would vanish ?). Kent was mainly working on some project he had with a private company, and in relation to our work, he was suggesting a number of variations to try. I was doing the computer graphics, and producing stimuli by the dozens. I would examine them, show them to Kent, and we would confront our reactions to these Hermann and Schrauf’s grids variants. On one occasion, making a variant in which Schrauf’s disks were smaller than usual, and the grid was rotated by 45 degrees, I noticed something strange. When I later gave the pile of stimuli to Kent, he almost jumped to the ceiling. He had made the same observation and had found it immediately extremely significant. That was it! In the extinction effect, black or white disks which are at the crossings of grey alleys simply disappear. The trick was to surround a white or black disk by a band of opposite colour so that the disk + the band had an average grey level which came close to that of the crossing alleys. When the eye is fixating a particular portion of the image containing disks, the disks are well perceived, but the disks which are away from the fixation point are not perceived at all, and instead one perceives continuous grey alleys which had been completed as in the case of alleys going across the blind spot. Our interpretation of the phenomenon was that objects at the periphery of the visual field do not catch attention, unless they have a sufficient local contrast with their surroundings. It would not be a question of central versus peripheral acuity, but a question of contrast thresholds.
After that, it was almost child play for me to produce the more effective version of the extinction effect, with a triangular lattice (see Figs. 1a and 1b), which was later published without difficulty in Perception [9]. A reviewer (who else than Nicholas Wade could it be ?) made an extremely pertinent link between our paper, and an earlier informal observation in a natural environment, made by Jeremiah Nelson, also published in Perception [10].
The Ninio-Stevens paper contained a number of original features. It included for the first time a version of Bergen’s scintillation effect which really worked on paper. This was obtained by introducing a "fuzziness gradient" in the image, so that people sensitive to the effect at different levels of fuzziness could all experience the phenomenon. Next, we mentioned the main geometrical constraint needed for the Hermann grid (the facilitatory effect of alignments of the crossings "may be mainly mediated through thre local lengthening of the branches of the crosses") and produced a quite sophisticated version of the Hermann grid (Fig. 1 of [9], top right panel) showing when the effect is, and when it is not produced. Four years after the publication of our article, which did not go unnoticed, there was in "Vision Research" a paper by two persons from the USA claiming that they had discovered a "new visual phenomenon", to which they gave the name of "blanking phenomenon". The paper was published during the dark era of the presidency of George Bush Jr in the United States, an era of arrogance towards the civilized world, an era of ruthless lies and appropriation of foreign property. McAnany and Levine's paper [11] came logically within the political context of this period. What is strange is that not only the authors dared to submit their paper in a scientific journal, but also that there were reviewers who tolerated the publication of this work. Very kindly, Akiyoshi Kitaoka noted in his website that he did not see in which way the "blanking phenomenon" was anything else than the previously published extinction effect.
FLASHING LINES
Figure 2 (pdf): flashing lines
Among the hundreds of variants of the Hermann grid I had produced during Kent Stevens stay at Ecole Normale Supérieure, I had tried distorted square lattices. How straight the alleys needed to be to produce the Hermann grid effect? I knew already that the alleys could be curved, (Fig. 6-6 in [2]) but what about more severe disruptions of continuity? I therefore tried several types of distorsions. So I used a matrix to produce periodic shape distorsions in the square lattice. In one of them, I noticed an effect which was much less impressive than the extinction effect, but which had perhaps more profound implications. I published it under the name of "flashing lines" [12].
Here, it seems that the brain picks up correctly some geometric feature in the figure: There are many white patches across which straight lines can go through. However, the lines cannot be white all the time, they are interrupted by black patches. So, it is like having dashed white lines, and the brain so to speak alternates between perceiving a continuous white line, or not perceiving it at all. The same phenomenon also exists in the opposite contrast.
ORIENTATION-DEPENDENT CONTRAST
Figure 3 (pdf): orientation-dependent contrast
I do not remember the precise motivations which made me design the stimuli giving rise to this effect. The basic observation was made on arrays of patches, each of which containing stripes at three different levels of grey and a single orientation. The figure contained two types of patches, differing only by the orientation of their stripes. While one family of patches, with stripes at one of the orientations seemed to be sharply contrasted, the patches having stripes at the other orientation had a washed out appearance. So, in this type of displays, the perceived contrast of a striped pattern depends upon the orientation of the stripes. People might immediately think that this is simply revealing some kind of astigmatism of the eyes. However, if this was the case, it would imply that a substantial fraction of the population have a form of astigmatism which is not usually revealed in current ophtalmological tests. I tend to believe that the orientation-dependent contrast is a real perceptual phenomenon. Old stimuli for the study of visual perception were often high contrast pure black and white figures. With the development of computer tools for graphic design, it is becoming very easy to produce high quality figures that use several intermediate levels of grey, in addition to black and white, and this boosted the emergence of a new generation of visual effects. The orientation-dependent contrast effect was described in [13]. The article also contained a figure showing a misorientation illusion, occurring when black or white lines went diagonally across a square tiling pattern, composed of striped tiles, in which the tiles had horizontal or vertical stripes in alternation (See section on geometrical visual illusions).
OUCHI VARIANTS
Figure 4 (jpeg): Ouchi variant with several orientations
Figure 5 (jpeg): Stereoscopic Ouchi variant
At ECVP 1994 in Eindhoven, I listened to a talk by Nicola Bruno in which he described the work he had done with Paola Bressan on the Ouchi illusion [14]. I was fascinated by the talk, and all the experiments which the authors had performed to characterize the effect. For incomprehensible reasons, the work could not be published at that time in a regular journal, but a much inferior work, by other authors, soon appeared in Vision Research in 1995 [15]. After this talk, I designed hundreds of variants on the Ouchi illusion. Two among the innovative variants were published in a popular science magazine [16], then in my book on illusions [2]. In one of the variants, I showed a section containing several distinct orientations, which was cohesively floating on a background also containing several distinct orientations (See Fig. 4). Presumably, it is not the orientation content of the figure and the ground that matter most, but how the orientations connect at the frontier between figure and ground. The other variant was a stereogram representing a truncated cone ( a lampshade shape) above a flat background. Both were represented with striped textures typical of Ouchi stimuli. The cone is perceived floating in 3D (Fig. 5). Whereas it was already known that the Ouchi effect worked in stereograms, this had been demonstrated for a flat figure over a flat ground. In my stimulus, the stereoscopic interpretation had to be carried out one step further, since a 3D shape had to be assigned to a figure in depth.
THE HULLA-HOOP ILLUSION
Figure 6 (pdf): Hulla-hoop illusion
I was looking for a variant of the Ouchi illusion which would work by rotation instead of working by translation. I produced the pattern in Fig. 6 which has interesting properties, whether one wishes to call it an Ouchi pattern or not. If this image is represented on a sheet of paper, and you submit the sheet to a rotating motion, without changing its orientation (the way you would move a cup of coffee, to dissolve the sugar in it) the inner circle, which is at the frontier between the striped patterns is seen to move with respect to the outer frontier of the image. If you move the sheet at the right speed, you can carry the inner circle in the rotation, and see it move as a hulla-hoop around the hips. The pattern appeared as Fig. 11.6 in [2]. More generally, when striped patterns are separated by a curve, and you move the figure, you can observe lagging motions of the curve. Another example (Fig. 7) was shown in Fig. 3 of [17] and Fig. 3a of [18].
GLIDING CIRCLES
Figure 8 (pdf): gliding circles
In the figures prepared to demonstrate the extinction effect (Fig. 4 and 5 in [9], and Fig.1a and 1b here), I included very small disks, to show that the extinction effect was not due to insufficient spatial resolution. I noticed that when the figure was translated back and forth in the direction of the numbered lines carrying the small disks, they were perceived as gliding back and forth between the two closest crossings. The effect was mentioned in the legend to Fig. 4 in [9]. It is shown here for its own sake in Fig. 8, and it was also published in [18] and in the Japanese edition of [2].
COLOUR-EMITTING WHEEL
Figure 9 (pdf): colour-emitting wheel
I do not remember the origin of this pattern. When the sheet on which this pattern is printed is moved in the same way as for the Hulla-hoop illusion (as for dissolving sugar in a cup of coffee), the various black, grey or white rings can be seen rotating like wheels, and a subjective colour may appear at the center. The pattern was published in Fig. 2 of [18] and in the Japanese version of [2]. For a long time I had been interested in producing an equivalent of the Benham top, which did not require a motor. This colour-emitting wheel might well be a step in this direction.
============================
REFERENCES
[1] Ninio, J. (1981) Random-curve stereograms: A flexible tool for the study of stereoscopic vision. Perception 10, 403-410.
[2] Ninio, J. (1998) La science des illusions. Odile Jacob, Paris. Also in English (The science of illusions, Cornell University Press; 2001), in German (Macht Schwarz schlank ?, Gustav Kiepenhauer, Leipzig), in Greek ((E epistimi ton psevdaithiseon, Katoptro, Athens, 2000) and Japanese (Shio-yo-sha, Tokyo, 2004).
[3] Stevens, K.A. and Brookes, A. (1987) Probing depth in monocular images. Biological Cybernetics 56, 355-366.
[4] Lulich, D.P. & Stevens, K.A. (1989) Differential contributions of circular and elongated spatial filters to the Café wall illusion. Biological Cybernetics 61, 427-435.
[5] Spillmann, L. (1994) The Hermann grid illusion: a tool for studying human receptive field organization. Perception 23, 691-708.
[6] Bergen, J.R. (1985) Hermann's grid: new and improved. Investigative Ophtalmology & Visual Science, Supplement 26, 280.
[7] Schrauf, M., Lingelbach, B and Wist, E.R. (1997) The scintillating grid illusion. Vision Research 37, 1033-1038.
[8] Baumgartner, G. (1960) Indirekte Grössenbestimmung der rezeptiven Felder der Retina beim Menschen mittels der Hermannschen Gittertäuschung. Pflüger Archiv für die gesamte Physiologie des Menschen und der Tiere 272, 21-22.
[9] Ninio, J. and Stevens, K. (2000) Variations on the Hermann grid: an extinction illusion. Perception 29, 1209-1217.
[10] Nelson, J.J. (1974) Motion sensitivity in peripheral vision. Perception 3, 151-152.
[11] McAnany, J.J. & Levine, M.W. (2004) The blanking phenomenon: a novel form of visual disappearance. Vision Research 44, 993-1001.
[12] Ninio, J. (2001) Flashing lines. Perception 30, 253-257.
[13] Ninio, J. (2002) Orientation-dependent contrast. Perception 31, 637-640.
[14] Bruno, N. and Bressan, P. (1994) Paradoxical motion in stationary patterns. Perception 23, supplement, 28.
[15] Hine,T.J., Cook, M. & Rogers, G.T. (1995) An illusion of relative motion dependent upon spatial frequency and orientation. Vision Research 35, 3093-3012.
[16] Ninio, J. (1996) Flottements. Pour la Science 223, 96-97.
[17] Ninio, J. (2000) Aspects of human shape perception. In Pattern formation in biology, vision and dynamics (A. Carbone, M. Gromov and P. Prusinkiewicz, eds). World Scientific, Singapore, pp. 365-381.
[18] Ninio, J. (2003) Faux mouvements. Pour la science, dossier hors-série 39 (Les illusions des sens, avril/juin 2003), 36-37.
