According to wikipedia, after images are “an optical illusion that refers to an image continuing to appear in one’s vision after the exposure to the original image has ceased”. The most common after image people experience is probably seeing dark spots after looking at a bright light, or seeing a natural coloured image after staring at an inverted colour version of the image.
Visual illusions are useful in that they tell us something about how the visual system works. Negative after images are thought to be the result of the light sensitive neurons in the retina adapting to an unchanging input. When they eyes are moved from the image to a blank page, the adapted neurons transmit a weak signal, but non-adapted neurons are still responsive, and send out a strong signal. In the case of the US flag above, the green stripes tire out the green receptors, producing an complimentary red after image. However, negative after images aren’t entirely driven by retinal activity.
Filling in: non-retinal afterimages
If after images were solely due to neurons in the retina adapting to a stimulus, negative after images could only consist of features present in the original stimulus. A study by Shimojo and colleagues demonstrated that stimuli that produce illusory contours like the Varin illusion (panel A below) can produce a standard negative after image (panel B, left) with the local features of the original image in tact, or after negative after images where the illusory contours of the implied shape are filled in (panel B, centre and left). As illusory contours are not in the original image they are probably the result of a process of perceptual adaptation in the cortex, rather than a process of physical adaptation in the retina. (The paper also shows further experiments that demonstrate that the process is the result of a global after image rather than the filling in of illusory contours in a local after image by manipulating the stimulus and the afterimage.)
Seeing edges that aren’t there
The Shimojo paper is a good demonstration of how what we ‘fill in’ what we see with what we expect. But if there is a stimulus present, and there is nothing to fill in, do after images still have an effect? Apparently so. Participants exposed to a particular polygon as a stimulus often reported an afterimage that was not the same shape as the original image, or blurred, but was instead a different, distinct polygon. A recent paper by Hiroyuki Ito investigated the phenomenon. Ito had noticed that participants who had adapted to hexagons often reported a circular after image, and vice versa. They offered two hypothesis – 1) the filling in hypothesis: the after images are the result of the background encroaching on the edges of the shape. This effect would be present on filled in shapes, but on unfilled shapes an encroaching background would erase the edges of the shape, rather than distorting the outline of the shape. 2) the line approximation hypothesis: the after images are the result of a cortical process of approximating a shape as a series of straight lines. This effect would produce hexagonal after images with circular stimuli, but hexagonal stimuli should retain their shape.
The results showed that when viewing unfilled circles, hexagonal afterimages predominated, suggesting that the filling in was not responsible for the after image. However, when viewing unfilled hexagons, both hexagonal and circular after images were typical, and circular after images were more common than after viewing a circle. When rotating shapes were presented, circular after images after viewing the hexagon were even more dominant.
The change in shape strongly suggests that the after image is the result of cortical processes. Primary visual cortex contains a significant number of “orientation selective” neurons who respond best to lines of a particular orientation. The line approximation hypothesis suggests that approximation of a curved line by these neurons, then their subsequent adaptation may produce the hexagonal after images. However, the circular after images produced by hexagonal stimuli are harder to explain. It may be that orientation selective neurons responding to the outline of the hexagon may adapt, resulting in a perceived shape that is a combination of all other orientations, which looks approximately like a circle. Alternatively, there may be some adaptation in part of the visual system dealing in more abstract curve/corner recognition system.
Inducing visual after images without a before image…
After images result from prolonged viewing of a stimulus, both for low level and for high level stimuli. The characteristics of the after image are usually complementary or opposite to those of the initial stimulus. But it is possible to induce a visual after image with no physical similarities to the original stimulus. When presented with an androgynous face, the perceived gender of the face can be manipulated by preceding it with an exaggerated male or female face – the androgynous face appears more feminine in contrast with a male face. However, Ghuman and colleagues were able to induce a gender after image on an androgynous face by preceding it with pictures of gendered human bodies without faces. Adaptation and after images from simple stimulus properties such as colour and movement have long been known. Recent work has shown that the phenomenon extends to higher level properties such as object shape and face properties. The research presented by Ghuman demonstrates that visual after images need not be related to stimulus properties – face after effects can result from non-face stimuli. The effect initially appears to be an after image driven by adaptation to the concept of a gender; however, the effect was not found when pictures of “gender connotative” objects were used (football helmets, purses), even though participants reported being aware that the objects were intended to be gender specific. The authors suggest that perhaps there is a distinction between viewing intrinsically gendered pictures (of bodies) and viewing pictures of items that are culturally gendered.
The study offers an interesting insight into the process of visual adaptation. It would be interesting to see what neurons are active during adaptation to non-face stimuli – is the adaptation the result of an activation of a distinct gender representation network, or are bodies and faces represented in the same place? On a more conceptual level, what other categories could be activated to produce after images? Is the distinction between inherent gender of bodies and abstract gender of culturally gendered items valid? And can we ever trust our vision?
Shimojo S, Kamitani Y, & Nishida S (2001). Afterimage of perceptually filled-in surface. Science, 293 (5535), 1677-80 PMID: 11533495
Ito H (2012). Cortical shape adaptation transforms a circle into a hexagon: a novel afterimage illusion. Psychological science, 23 (2), 126-32 PMID: 22207643
Ghuman, AS, McDaniel, JR, & Martin, A (2010). Face Adaptation Without a Face Current Biology, 20 (1), 32-36 DOI: 10.1016/j.cub.2009.10.077