Can an afterimage be filled in from stereoscopic cues?

Elian Malkin

Fixating on a point such that the cones of the retina are continually exposed to a given color induces adaptation - the sensitivity of the cones to the color is reduced. When the observer is then shown a white screen, an afterimage is perceived in the complementary color. The paper Filling-in afterimage colors between the lines (van Lier R, Vergeer M, Anstis S.) demonstrates that the color of the afterimage depends on contours shown on the blank screen. The afterimage of two different colors will be perceived strongly as one color if a contour is drawn around the edges of that color. Moreover, the regions within the contour that were not adapted to color are filled in, creating a percept of a uniform color fill of the contour.

Link to original illusion.

I wanted to explore whether the same effect could be elicited by contours created by stereoscopic effects. My illusion uses the same star-like shapes as the original illusion, with the same colors. As before, two identical images are presented side by side, however, rather than fixating on a point between them, the observer diverges or converges their eyes to create a single image formed from the combination of the two adjacent images - a stereogram. You should see one sharp image in the center with two peripheral images on either side. Once the new focus plane is acquired, the observer should fixate on the yellow dot. The image will then switch to the same shape in white on a textured background. The two different star shapes, now in white, will alternate in moving stereoscopically towards the observer. This is done by shifting the corresponding shape on the right slightly to one side. The textured background serves only to provide background cues for the stereoscopic effect. Both white star shapes are presented simultaneously, with the only difference being the perceived plane on which they lie. This allows us to determine whether stereoscopic cues produce the same effect as contour cues in the original illusion. Is the percept of the afterimage altered depending on which of the two star shapes is at the forefront? Does the afterimage change as the shapes move towards and away from the observer?

Several loops of the video should be viewed for the strongest afterimage effect, just as in the original.
If you are having difficulty forming the stereogram, try zooming out on the GIF - the smaller the shapes, the smaller the divergence/convergence of your eyes should be.
I have provided a GIF for those who converge and those who diverge. The only difference is the direction in which the shapes are shifted, which alters whether the shapes move in or out of the plane. The convention for viewing stereograms tends to be for shapes to move towards you, out of the plane.

Diverge:


Converge:




Citations:
van Lier R, Vergeer M, Anstis S. Filling-in afterimage colors between the lines. Current Biology : CB. 2009 Apr;19(8):R323-4. DOI: 10.1016/j.cub.2009.03.010.

Comments


Elian Malkin

c) The reports suggest that the stereoscopic cues do affect the color of the afterimage, though not as much as contours in the original illusion. While the contours were very effective at highlighting a particular color and making that color fill the whole shape, the stereoscopic contour in my illusion is a somewhat weaker cue to accept the afterimage as the color of a shape.

d) Because both star shapes are overlaid on a dark background, the two afterimage colors can both be identified quite easily as the black background squares act as a border around each color. In the original illusion, the only border present was the contour that enclosed only one of the colors, hence making it stand out against the other color falling on a purely white background. One change I would make to my illusion would be to find a textured background that has a light enough average luminance to allow the perception of afterimages (works best on a white background), but has sufficiently dense features to create a stereoscopic effect. I could shift a star-shaped portion of the texture to create the same stereoscopic cues, but this time without any dark borders around either shape. This would more clearly isolate the stereoscopic effect from the contour effect.

Michael Anoke

a. As a subject, I was able to perceive an after image, though it is more difficult to perceive than the original illusion is. It was a very nice illusion! (I love stereograms)

b. I think the illusion did help answer the question, but I believe what Jeremy said was correct. my only concern was that when I was watching the gif, the yellow dot seemed to move with each shift. Other than that, I think the illusion is great.

Maddie C

I tried looking at this 2D at first (looking at the right side of the image only). I didn't get a strong sense of colour change. Then I looked at it 3D and I got was a much stronger color change. For me, the very inside of the star remains white, which is interesting -- the colour doesn't seem to spread as much to the middle of the shape as it does in the original paper (with outlines).

Malinda

Cool illusion! I definitely see that moving between the planes changes the color of the afterimage.

Jeremy Ma

a) As an observer I really appreciate that there are both converging and diverging options! However, the instructions could be slightly more clear though, since it took me a while to understand what I am looking for. About the illusion, I do indeed see an after image pretty vividly, and the moving between planes do indeed change the color of the after image but it is not as obvious as the 2D scenario.

b) I think one thing is that since you are only shifting one image to the right, this makes it seem like the stereoscopic image is moving to the right too, while in reality you should've moved both images away from the center half as much to preserve the location of the star. Other than that, very fascinating illusion! And really shows the association between color and shape is pretty high level.