Afterimage

Abstract

An afterimage is a form of visual perception: a visual sensation that persists or emerges after its physical stimulus has been removed. The classic case is the negative afterimage, in which a steadily fixated colour is followed, on a blank field, by a ghost in the complementary hue. Afterimages were long treated as a purely retinal residue of bleached photopigment and adapted receptors, and much of that story holds. Yet the modern account is layered: the signal begins in cone-specific retinal adaptation, but its perceived size scales with distance rather than retinal extent, its colour is filled in within contours presented after adaptation, and its visibility is modulated by binocular grouping, contextual edges, and even the vividness of mental imagery. The afterimage is therefore a small window onto how the visual system builds a percept from an adapting signal.

Keywords: afterimage, negative afterimage, opponent process, neural adaptation, filling-in

An afterimage is a visual percept that outlasts, or arises only after, the removal of the stimulus that produced it. Prolonged fixation of a bright or coloured region alters the local sensitivity of the visual system, so that a subsequently viewed blank surface is not seen as uniform: it carries a patterned residue in the shape of the original stimulus. In the most familiar demonstration, steady inspection of a saturated colour is followed by an afterimage in the opponent, complementary hue — red yields cyan-green, blue yields yellow (Hurvich & Jameson, 1957). Afterimages matter to cognitive psychology because they expose adaptation, the process by which sensory systems continuously rescale themselves to the prevailing input, and because the humble afterimage turns out to be shaped at several levels of the visual hierarchy rather than at the retina alone (Zaidi et al., 2012).

Negative and Positive Afterimages

Two broad classes are distinguished by how the afterimage relates to the original. A negative afterimage reverses the stimulus: bright regions appear dark, and each colour is replaced by its complement. It is the enduring, easily produced kind and follows tens of seconds of steady fixation. A positive afterimage preserves the polarity of the stimulus — light where the stimulus was light — and is brief, most readily seen after a short, intense flash in a dark surround, as when a camera flash leaves a lingering bright blob. Negative afterimages dominate the experimental literature because they are stable enough to measure, and because their complementary colour makes them a direct probe of the opponent organisation of colour vision (Hurvich & Jameson, 1957).

The afterimage is not a static residue but a signal that behaves like a real stimulus in its own right. Its brightness can be matched against an equivalent physical background, and when that is done the afterimage is found to add to, and interact with, ongoing retinal signals in the same currency as light (Barlow & Sparrock, 1964). This equivalence is the foundation for treating the afterimage as an adapting signal that the rest of the visual system must then interpret.

Figure 1

Opponent Channels and the Complementary Afterimage

The Opponent-Process Account

Why a negative afterimage takes the complementary colour is explained by the opponent organisation of colour vision. In opponent-process theory, colour is coded not by three independent cone signals but by their differences, along a red-green channel and a blue-yellow channel, with a separate light-dark channel (Hurvich & Jameson, 1957). Prolonged exposure to red drives the red-green channel toward its red extreme and reduces its responsiveness there; when the stimulus is removed, the channel rebounds past neutral toward green, and a neutral field is seen as tinged green. Because each hue has a fixed opponent partner, the afterimage colour is predictable from the adapting colour, which is precisely what makes the negative afterimage a clean demonstration of opponency rather than a curiosity. Recent work continues to exploit this logic, using adaptation and afterimages to probe how the visual system fixes its neutral point, the balance of channels that defines perceptual white (Anstis et al., 2024).

Receptor Bleaching and Neural Adaptation

The oldest account located the afterimage in the photoreceptors: intense light bleaches photopigment, leaving the exposed patch of retina temporarily less sensitive, so that a uniform field stimulates the surrounding, unbleached retina more strongly and the bleached region appears dark. Bleaching is real and contributes, and the afterimage's brightness can be tracked precisely as an equivalent background across the course of dark adaptation (Barlow & Sparrock, 1964). But bleaching alone cannot be the whole story. Afterimages can be generated by adapting stimuli far too dim to bleach appreciable pigment, and they can last far longer than the photochemical time constant allows, which points to adaptation in the neural machinery beyond the receptors (Virsu & Laurinen, 1977).

Locating the signal more precisely, the negative colour afterimage has been traced substantially to cone-specific and post-receptoral adaptation within the retina itself, on the basis of its time course and its selectivity for the adapted cone classes (Zaidi et al., 2012). The picture that emerges is not receptor versus neuron but a layered one: a peripheral, largely retinal generator produces an adapting signal that later stages then interpret and can override.

Filling-In and Surface Representation

If the afterimage were simply a fixed retinal residue, its appearance would be dictated entirely by what was adapted. It is not. When an outline is presented after adaptation, the afterimage colour is filled in within that contour, so a single adapting pattern can yield differently shaped, differently coloured afterimages depending on the outline shown afterward (van Lier et al., 2009). The afterimage colour behaves like a fill that flows up to the nearest bounding edge, a construction rather than a stamp. The same conclusion follows from surface-level manipulations: the afterimage of a region tracks its perceptually filled-in surface rather than the raw retinal stimulus, so what is adapted at the level of a completed surface representation determines the subsequent afterimage (Shimojo et al., 2001). Filling-in places part of the afterimage percept at a stage where surfaces, not just point-by-point light, are represented.

Emmert's Law and Perceived Size

An afterimage occupies a fixed patch of retina, so its retinal angle does not change. Yet its perceived size is not fixed: projected onto a near surface it looks small, and onto a far surface it looks large. Emmert's law states the regularity — the perceived size of an afterimage grows in proportion to the perceived distance of the background it is cast upon, because the visual system scales a constant retinal angle by distance to recover physical size, exactly as it does in size constancy for real objects (Sperandio et al., 2012). The striking modern finding is neural: retinotopic activity in primary visual cortex tracks the perceived size of an afterimage, expanding and contracting with the distance of the surface, even though the retinal stimulation is unchanged (Sperandio et al., 2012). A signal that begins as fixed retinal adaptation is thus re-represented in cortex according to how large the observer perceives it to be.

Cortical and Top-Down Modulation

Several phenomena show that the afterimage percept is shaped well beyond the retina. That adaptation itself reaches cortex is shown by the McCollough effect, an orientation-contingent colour aftereffect in which staring at alternating coloured gratings makes a later black-and-white grating look tinged with the complementary colour, but only at the adapted orientation — a pattern-contingent, plainly cortical form of adaptation distinct from a simple retinal afterimage (McCollough, 1965). Beyond adaptation, the afterimage's perceived strength depends on the context of edges around it: a matching contour makes an otherwise weak afterimage credible and vivid, and afterimages are modulated by contextual edges more strongly than real stimuli of the same contrast are (Powell et al., 2012). Binocular interactions matter too — afterimages participate in interocular grouping, in which fragments presented to the two eyes are combined into a coherent afterimage, a feat that requires binocular, cortical machinery (Dong et al., 2017). The afterimage signal is not perceptually inert either: a negative afterimage summates with a subsequently viewed real image, lowering the contrast threshold for detecting it, which shows the afterimage being processed alongside genuine input rather than merely subtracted from it (Kingdom et al., 2020). Attention modulates the afterimage as well, and in a way that dissociates two processes usually confounded: attending to the adapting stimulus during induction weakens the resulting afterimage, whereas making that same stimulus more consciously visible strengthens it, so attention and consciousness exert opposing effects on afterimage strength (van Boxtel et al., 2010). Finally, individual differences reach up to imagery: the vividness of a person's afterimage percept correlates with the vividness of their voluntary mental imagery, linking the afterimage to the same top-down visual processes that support imagination (Kronemer et al., 2024).

Worked Example

Consider an observer who fixates a small bright cross that bleaches a patch of retina subtending a visual angle of 2.0°. The afterimage is then projected, in turn, onto a wall 1.0 m away and a wall 3.0 m away. Because the bleached patch is fixed on the retina, the retinal angle is the same, 2.0°, on both walls; only the perceived distance differs. The physical size an afterimage appears to have on a surface follows the geometry of the visual angle, S = 2D·tan(θ/2), where D is distance and θ is the retinal angle. With θ = 2.0°, tan(θ/2) = tan(1.0°) = 0.017455. On the near wall, S = 2 × 1.0 × 0.017455 = 0.0349 m, about 3.5 cm. On the far wall, S = 2 × 3.0 × 0.017455 = 0.1047 m, about 10.5 cm. Tripling the distance triples the perceived size, the ratio predicted by Emmert's law — and, on the neural evidence, the retinotopic extent of the afterimage's representation in primary visual cortex expands by the same factor even though the retinal image never changed (Sperandio et al., 2012).

Discussion

The afterimage repays close study because it dismantles a tidy but wrong intuition — that a percept without a present stimulus must be a passive retinal leftover. The generator is indeed largely peripheral: cone-specific adaptation and photopigment dynamics supply the signal (Zaidi et al., 2012; Barlow & Sparrock, 1964). But everything downstream — the complementary colour set by opponent channels, the size set by perceived distance, the shape set by contours imposed after the fact, the credibility set by surrounding edges, the visibility set by binocular grouping and even imagery — is the work of later visual processing (Hurvich & Jameson, 1957; Sperandio et al., 2012; van Lier et al., 2009; Powell et al., 2012; Dong et al., 2017; Kronemer et al., 2024). The afterimage is a rare case in which an adapting signal of known, low-level origin can be watched as it is progressively reinterpreted, which is why it remains a useful tool for asking where in the visual hierarchy a given perceptual attribute is fixed. It also sharpens a conceptual distinction that is easy to blur: an afterimage, a residue of adaptation to a static stimulus, is not the same thing as a perceptual aftereffect such as the motion aftereffect, in which adaptation to one feature biases the perception of a later, different stimulus (Wade, 1994).

Common Misconceptions

An afterimage is just bleached pigment in the eye.

Bleaching contributes, but afterimages arise from adapting stimuli too dim to bleach much pigment and can outlast the photochemistry, which points to neural adaptation beyond the receptors (Virsu & Laurinen, 1977). Their colour and size are then set by cortical processing, not by the retina alone (Sperandio et al., 2012).

The afterimage has a fixed size.

The retinal patch is fixed, but the perceived size grows in proportion to the distance of the surface the afterimage is cast on, by Emmert's law, and primary visual cortex tracks that perceived size rather than the constant retinal angle (Sperandio et al., 2012).

The afterimage colour is locked in at the moment of adaptation.

A contour presented after adaptation captures and reshapes the afterimage colour, filling it in between the lines, so the same adapting stimulus can yield different afterimages depending on what outline is shown next (van Lier et al., 2009).

Glossary

Adaptation.

The continuous rescaling of a sensory system's responsiveness to the prevailing level and pattern of stimulation.

Afterimage.

A visual sensation that persists or emerges after the stimulus that produced it has been removed.

Complementary colour.

The opponent partner of a hue, such as green for red, in which a negative colour afterimage appears.

Cone.

A retinal photoreceptor mediating daylight and colour vision, of three spectral classes, whose selective adaptation shapes colour afterimages.

Dark adaptation.

The gradual recovery of visual sensitivity in darkness, against which an afterimage can be measured as an equivalent background.

Emmert's law.

The rule that the perceived size of an afterimage is proportional to the perceived distance of the surface it is projected upon.

Filling-in.

The perceptual completion of a region's colour or brightness up to its bounding contours, which shapes the appearance of an afterimage.

Interocular grouping.

The binocular combination of fragments presented to the two eyes into a single coherent percept, shown for afterimages.

McCollough effect.

An orientation-contingent colour aftereffect in which adaptation ties an illusory colour to a grating's orientation, a pattern-contingent, cortical form of adaptation.

Negative afterimage.

An afterimage that reverses the stimulus, appearing dark where it was bright and in the complementary hue.

Opponent process.

The coding of colour along antagonistic red-green and blue-yellow channels, which predicts the complementary colour of a negative afterimage.

Perceptual aftereffect.

A bias in the perception of one stimulus caused by prior adaptation to another, such as the motion aftereffect, distinct from a static afterimage.

Photopigment bleaching.

The light-driven breakdown of photoreceptor pigment that lowers local sensitivity and contributes to, but does not fully explain, afterimages.

Positive afterimage.

A brief afterimage that preserves the polarity of the stimulus, seen after a short intense flash in a dark surround.

Retinal angle.

The visual angle subtended by a stimulus at the eye, which for an afterimage stays constant regardless of projection distance.

Key Researchers

Stuart Anstis (b. 1934). Emeritus Professor of Psychology at the University of California, San Diego; he showed how afterimage colour is captured and filled in between contours defined after adaptation, treating the afterimage as an edge-bounded construction. Faculty Page - Wikipedia

Rob van Lier. Professor at the Donders Institute for Brain, Cognition and Behaviour, Radboud University; he demonstrated that afterimage colours are filled in within contours presented after adaptation, so one adapting pattern yields different afterimages depending on the outline shown next. Faculty Page - ORCID

Shinsuke Shimojo. Professor of Biology at the California Institute of Technology; he showed that the afterimage of a region follows its perceptually filled-in surface rather than the raw retinal stimulus, placing part of the afterimage percept at a surface-based stage. Faculty Page - ORCID

Qasim Zaidi. Distinguished Professor at the State University of New York College of Optometry; he located the neural origin of colour afterimages substantially in cone-specific and post-receptoral retinal adaptation, using their time course and cone selectivity. Faculty Page - ORCID

Frequently Asked Questions

Why does a negative afterimage appear in the opposite colour?
Colour is coded along opponent red-green and blue-yellow channels; prolonged exposure to one colour fatigues its side of the channel, so on a neutral field the channel rebounds toward the complementary hue (Hurvich & Jameson, 1957).

Is an afterimage caused only by bleaching of pigment in the eye?
No. Bleaching contributes, but afterimages appear after stimuli too dim to bleach much pigment and outlast the photochemistry, implicating neural adaptation beyond the receptors (Virsu & Laurinen, 1977).

Where in the visual system does a colour afterimage originate?
Its origin is substantially cone-specific and post-receptoral adaptation within the retina, inferred from the afterimage's time course and its selectivity for the adapted cone classes (Zaidi et al., 2012).

Why does an afterimage look bigger on a distant wall?
By Emmert's law the perceived size scales with the perceived distance of the surface, since the visual system rescales a constant retinal angle by distance, and primary visual cortex tracks that perceived size (Sperandio et al., 2012).

Can the shape of an afterimage change after the stimulus is gone?
Yes. A contour shown after adaptation captures the afterimage colour and fills it in between the lines, so the same adapting stimulus can produce differently shaped afterimages (van Lier et al., 2009).

Does the surrounding scene affect how strong an afterimage looks?
Strongly. A matching set of contextual edges makes an afterimage more credible and vivid, and afterimages are modulated by such edges more than real stimuli of the same contrast are (Powell et al., 2012).

Is an afterimage the same as a motion aftereffect?
No. An afterimage is a residue of adaptation to a static stimulus, whereas a perceptual aftereffect such as the motion aftereffect biases the perception of a different, later stimulus (Wade, 1994).

Do afterimages differ between people?
Yes. The vividness of a person's afterimage percept correlates with the vividness of their voluntary mental imagery, tying afterimages to top-down visual processes (Kronemer et al., 2024).

References