Visual Space: How the Seen World Is Laid Out

Visual space is the space we experience when we look: the arrangement of things as they appear to an observer, near and far, left and right, up and down. It is the perceptual counterpart of the physical space in which objects actually sit. The two are closely related, since we navigate the world successfully by sight, but they are not the same. The Wikipedia article on the subject puts the contrast sharply: visual space "can not be measured with rulers." You can measure the distance between two trees with a tape. You cannot run a tape through your experience of the trees. Yet that experience has structure: things in it are ordered, some appear farther than others, some lines look parallel and others converge.

Questions about visual space have occupied philosophers since antiquity and became experimental questions in the nineteenth century. What is the geometry of the space we see? Is it Euclidean, like physical space at everyday scales, or something else? How does a brain that receives two flat, upside-down images on the retinas build a three-dimensional world? Is our sense of visual distance innate or learned? This article covers the concept, its history in philosophy and science, the main experimental findings, the neural basis, and some common misunderstandings.

Three spaces that are easy to confuse

Physical space, visual field and visual space

The Wikipedia article draws a careful distinction among three things.

The article notes that German usage of the word Sehraum ("seeing space") blurred the distinction between visual field and visual space, and that when Ewald Hering and his followers used the term, they meant visual space.

Percepts as an ordered ensemble

The things located in visual space are percepts, the conscious counterparts of physical objects. The Wikipedia article describes percepts as forming an ordered ensemble: they stand in relations of nearer and farther, above and below, between and beside. Visual space is the structure of those relations.

Carnap's three kinds of space

The philosopher Rudolf Carnap, in his 1922 dissertation "Der Raum" (Space), published as a supplement to Kantstudien, distinguished three kinds of space, a distinction the Wikipedia article uses to frame the topic:

Visual space is a kind of perceptual space. Carnap's distinction makes clear why the question "what is the geometry of visual space?" is meaningful. Mathematicians had shown by Carnap's time that many consistent geometries exist (formal space), physics had shown that the geometry of physical space was an empirical question, and so it became natural to ask, separately, what geometry best describes perceptual space.

Philosophical history

Ancient and medieval optics

The study of how we see space began with optics. The Wikipedia article on the moon illusion records that the illusion that the Moon looks larger near the horizon was discussed by Aristotle around 330 BC and by Ptolemy in the second century, who attributed it to refraction in the Almagest and to apparent distance in his Optics.

The Arab scholar Ibn al-Haytham (known in Latin as Alhazen), in his Book of Optics written between 1011 and 1022, gave an account of the moon illusion based on apparent distance. He argued that judging distance requires a chain of intervening objects. Across the ground to the horizon, there are many such objects; straight up, there are none. So the overhead Moon seems nearer, and the horizon Moon farther and therefore larger. The Wikipedia article on unconscious inference credits Ibn al-Haytham with elaborate frameworks around 1030 for the idea that perception involves inference, a precursor of later theories. The Wikipedia article on the horopter notes that he also observed that objects on a horizontal line through the point of fixation appear single, an early observation about binocular vision.

Descartes, Molyneux and Locke

René Descartes, in 1637, contributed to the idea that the mind reasons from visual cues to distance, according to the Wikipedia article on unconscious inference.

The question that most sharply divided early modern philosophers was posed by the Irish scientist William Molyneux in a letter to John Locke dated 7 July 1688. According to the Stanford Encyclopedia of Philosophy, Molyneux asked: if a man born blind, who can tell a globe from a cube by touch, were suddenly made to see, could he tell them apart by sight alone? Locke did not reply at the time. Molyneux sent a simplified version in 1693, and Locke included it in the second edition of his Essay Concerning Human Understanding in 1694. Both answered no: the man would lack the experience linking how shapes feel to how they look. Locke wrote that he "would not be able with certainty to say, which was the Globe, which the Cube" at first sight.

Gottfried Wilhelm Leibniz, writing in 1704, answered yes, while arguing that the newly sighted person should be allowed time for sight to recover.

Berkeley's New Theory of Vision

George Berkeley's An Essay Towards a New Theory of Vision, published in 1709, gave the most radical answer. Berkeley argued that the ideas of sight and the ideas of touch have no similarity at all, and that distance is not immediately seen. The Stanford Encyclopedia notes that he conjectured that first sight would lack spatial extension, so that objects would seem to be in the eye. On Berkeley's view, visual distance is learned by associating visual signs with tactile and motor experience.

Cheselden's case

In 1728 the surgeon William Cheselden reported on a young man born blind whose cataracts he had removed. According to the Stanford Encyclopedia, Cheselden wrote that the patient "thought all objects whatever touched his eyes (as he expressed it) as what he felt did his skin." This seemed to support Berkeley, but writers including Voltaire and Condillac found the case ambiguous.

Reid's geometry of visibles

Thomas Reid, in his Inquiry into the Human Mind (1764) and Essays on the Intellectual Powers of Man (1785), developed Berkeley's distinction between visible and tangible figure. According to the Stanford Encyclopedia's entry on Reid, he argued that visible figures are geometrically equivalent to shapes projected onto the surface of a sphere centered on the eye. He also held that the variation of visible figure with distance and angle is systematic, not arbitrary: a table's visible figure "can be determined demonstratively by the rules of geometry and perspective." Whether Reid thereby gave the first recorded non-Euclidean geometry is debated; the encyclopedia entry presents it as an open question rather than a settled claim. The geometry of a sphere's surface is non-Euclidean, so the suggestion is natural, but Reid himself also tried to derive Euclid's parallel postulate.

Kant

Immanuel Kant's Critique of Pure Reason (1781) treated space as a form of intuition, a structure the mind brings to experience rather than something learned from it. The Wikipedia article on visual space links Carnap's perceptual space to Kant's Anschauungen. Kant also discussed the moon illusion, according to the Wikipedia article on that topic. Kant's view raised the question of whether the geometry of experienced space must be Euclidean. The later discovery of consistent non-Euclidean geometries made that question empirical.

Helmholtz and unconscious inference

Hermann von Helmholtz's Handbuch der physiologischen Optik (Handbook of Physiological Optics) appeared between 1856 and 1867. In it he coined the term "unconscious inference" in 1867, according to the Wikipedia article on the concept. Helmholtz proposed that perception results from involuntary, reflex-like inferences from sensory signals to their probable causes. Because these inferences never reach consciousness, reasoning cannot undo them; their results strike us as "a foreign and overpowering force of nature." Visual illusions, on this view, occur when the usual inferences are applied in unusual circumstances.

Mach, Cassirer and the twentieth century

The physicist and philosopher Ernst Mach published an analysis titled "On Physiological as Distinguished from Geometrical Space" in 1901, later collected in his 1906 book Space and Geometry. He argued that physiological space differs from geometric space: it is not uniform in all directions, and it is neither infinite nor unbounded. The philosopher Ernst Cassirer, in a 1944 paper on "The Concept of Group and the Theory of Perception," explored how the mathematical idea of a group of transformations applies to perception.

Molyneux's question tested

In 2011 Richard Held and colleagues reported a test of Molyneux's question in Nature Neuroscience. According to the Stanford Encyclopedia, five congenitally blind subjects whose sight was restored by cataract surgery passed vision-to-vision and touch-to-touch shape matching but failed the touch-to-vision test, performing "barely better than if the subjects had guessed." Three subjects succeeded when retested after a few days of real-world experience. The result supports Molyneux's and Locke's "no," with the notable twist that the link between touch and sight appears to be learned quickly.

Historical summary

Measuring visual space: psychophysics

Fechner's program

Philosophers studied visual space by introspection and argument. Psychophysics, founded by Gustav Fechner in the nineteenth century, sought experimental methods. Fechner defined psychophysics as the study of the functional relationship between the mental and material worlds. The Wikipedia article on visual space notes that he also recognized an intermediate step, the nervous system's response to stimuli, which he called inner psychophysics, as distinct from the outer relationship between stimulus and experience.

How you can measure something that cannot be measured with rulers

The trick is to ask observers to make perceptual judgments and then measure the physical arrangement that satisfies those judgments. The Wikipedia article gives an example. An observer adjusts two lines until they appear parallel. The experimenter then measures the lines' actual positions in physical space. If perceived parallels are physically parallel, that is one result; if they physically converge or diverge, the deviation reveals something about the structure of visual space. The article notes that this and similar methods have been used with human observers and with trained animals.

Many other judgments can be used the same way: setting points to appear equally spaced, setting a point to appear halfway between two others, setting a line to appear straight, or setting points to appear on a circle around the observer. Each produces a physical configuration that corresponds to a perceptual one, and comparing the two maps visual space onto physical space.

What is the geometry of visual space?

Luneburg's hyperbolic model

The most famous attempt to give visual space a definite geometry came from the mathematician Rudolf Luneburg (1903 to 1949), a German-born professor of mathematics and optics at the Dartmouth College Eye Institute. His 1947 book Mathematical Analysis of Binocular Vision, published by Princeton University Press, concluded from psychophysical data that binocular visual space is hyperbolic, a non-Euclidean geometry with constant negative curvature. According to the Wikipedia article on Luneburg, his analysis assumed that the angle of convergence between the eyes provides a constant measure of distance.

Hyperbolic geometry has striking features. Through a point not on a line there are infinitely many lines that do not meet it. The angles of a triangle add up to less than 180 degrees. And a hyperbolic plane can be mapped so that its infinite extent fits within a bounded region. The Wikipedia article on visual space notes that Luneburg's model turns infinity into a "dome," which he identified with the sky. The everyday impression that the sky is a bowl over our heads, with a finite distance to the horizon, fits that picture.

Did the model hold up?

The Wikipedia article states that Luneburg drew his conclusions from limited psychophysical data and that follow-up experiments mostly did not support his proposal. Later investigators, such as J. M. Foley in a 1964 paper in the Journal of the Optical Society of America on the "Desarguesian property in visual space," tested specific geometric properties that a space of constant curvature should have. A comprehensive review is Mark Wagner's 2006 book The Geometries of Visual Space, cited in the Wikipedia article; its plural title reflects the field's conclusion that no single geometry fits.

The main obstacle is variability. The Wikipedia article notes that geometric conclusions depend on the observer, the day and the context, and that any mathematical model of visual space therefore has limited validity. Results obtained in a dark room with a few points of light differ from those obtained in a furnished room or outdoors. Results for near space differ from far space. Results for different tasks, such as judging parallels versus judging distances, do not always fit the same geometry.

Why accurate navigation is still possible

The variability might suggest that visual space is a mess. It is not, at least for practical purposes. The Wikipedia article stresses that visual location closely matches physical location, which allows effective navigation. Geometrical-optical illusions are common but minor in effect. We reach for cups, catch balls and walk through doorways without measurable difficulty. Deviations from Euclidean structure show up mainly in carefully designed experiments, in reduced viewing conditions, and over large distances.

The article also notes that visual space agnosia is a recognized neurological condition, in which patients have difficulty perceiving spatial relations even though they can see. Its existence shows that the construction of visual space is a specific brain function that can fail independently of basic vision.

How depth gets into visual space

The ill-posed problem

Each retina receives a two-dimensional image. Many different three-dimensional scenes could produce the same image. The Wikipedia article calls the conversion of a two-dimensional retinal image into a three-dimensional percept mathematically "ill-posed," meaning it has no unique solution without additional assumptions. Humans nonetheless do it effectively, by combining many sources of information called depth cues.

Monocular cues

The Wikipedia article on depth perception lists cues available to a single eye:

Binocular cues

The depth perception article notes that accommodation is effective only at distances below about 2 meters and convergence only below about 10 meters, and that only convergence, accommodation and familiar size give absolute distance; the other cues give only relative depth.

Worked example: why convergence fails at a distance

The following calculation shows why convergence is useful only for near objects. It uses a hypothetical interocular separation of 6.3 centimeters (a round figure for illustration, not a population statistic). When both eyes fixate a point straight ahead at distance D, the angle between the lines of sight is 2 × arctan(0.0315 / D), with D in meters.

Between 0.5 and 2 meters the angle changes by more than 5 degrees, a large signal. Between 10 and 50 meters it changes by less than 0.3 degrees, a tiny one. Doubling distance roughly halves the angle, so equal changes in angle correspond to ever larger changes in distance. This compression is consistent with the observation that convergence is effective only at short range, and it hints at why the geometry of visual space might differ between near and far.

The horopter and corresponding points

Binocular vision depends on the idea of corresponding retinal points: pairs of points, one in each eye, that see in the same visual direction. The horopter is the set of points in space whose images fall on corresponding points. According to the Wikipedia article on the horopter, the term was introduced in 1613 by Franciscus Aguilonius. In 1818 Gerhard Vieth argued from Euclidean geometry that the horopter is a circle passing through the fixation point and the eyes' optical centers; Johannes Müller reached a similar conclusion, and the theoretical curve is called the Vieth-Müller circle. Charles Wheatstone found in 1838 that the empirical horopter differs from the theoretical one. Peter Ludvig Panum proposed in 1858 that each retinal point is surrounded by a small region, now called Panum's fusional area, within which images from the two eyes fuse into single vision.

The measured horizontal horopter deviates systematically from the Vieth-Müller circle, a pattern called the Hering-Hillebrand deviation: it is flatter than geometry predicts at short distances and becomes convex at longer ones. Recent researchers propose that these deviations reflect the visual system's adaptation to the statistics of natural scenes.

Stereopsis

Wheatstone's 1838 paper in the Philosophical Transactions of the Royal Society introduced the stereoscope and showed that two flat images differing only in horizontal disparity produce a vivid impression of depth. The Wikipedia article on stereopsis records that Horace Barlow, Colin Blakemore and John Pettigrew in 1967 found neurons in the cat visual cortex tuned to disparity, establishing a neural basis for stereopsis; David Hubel and Torsten Wiesel, initially skeptical, found similar neurons in monkeys by 1970. Béla Julesz's random-dot stereograms, presented in his 1971 book Foundations of Cyclopean Perception, showed that stereoscopic depth can be seen even when neither eye's image contains any recognizable shape, proving that depth can be computed before objects are recognized.

The Wikipedia article on visual space notes an important limitation: stereopsis gives relative depth directly, but how it relates to distance in the other two dimensions is uncertain. Seeing that one thing is in front of another by some amount does not by itself tell you how far away either is.

Size, distance and illusions

Emmert's law and size-distance invariance

In 1881 Emil Emmert observed that an afterimage appears larger when projected onto a more distant surface. Emmert's law states that objects casting retinal images of the same size appear larger when they appear farther away. It is a special case of the size-distance invariance hypothesis, which proposes that perceived size and perceived distance are linked through the visual angle.

Worked example (geometry only, hypothetical numbers): an afterimage occupies a fixed visual angle, say 2 degrees. If it is seen on a wall 1 meter away, the physical size that subtends 2 degrees at 1 meter is 2 × 1 × tan(1 degree) ≈ 0.035 meters, about 3.5 centimeters. On a wall 3 meters away, the same angle corresponds to about 0.105 meters, about 10.5 centimeters. Geometry predicts a threefold increase in apparent size. The Wikipedia article on Emmert's law notes that the actual perceived growth is smaller than geometry predicts, which is itself evidence that visual space is not a simple scaled copy of physical space.

The moon illusion

The Moon's angular size barely changes as it rises. The Wikipedia article on the moon illusion notes that a pebble held at arm's length covers it equally well at the horizon and overhead, and that the Moon is about 1.5 percent farther away at the horizon, so its image is actually slightly smaller there. Its angular diameter varies over the lunar orbit from about 29.43 arcminutes at apogee to 33.5 at perigee. Yet the horizon Moon typically looks about 50 percent larger in diameter, according to Ross and Plug (2002).

Worked example (geometry only): at a mid-range angular diameter of 31 arcminutes, the object that would just cover the Moon at 70 centimeters from the eye is 2 × 0.7 × tan(15.5 arcminutes) ≈ 0.0063 meters, about 6.3 millimeters, roughly the size of a pea or small pebble. That tiny size is the same at the horizon and at the zenith, which is why the pebble test works.

Explanations remain disputed. The apparent-distance theory, going back to Ibn al-Haytham, holds that the sky appears as a flattened dome, so the horizon seems farther than the zenith, and by size-distance reasoning the same angular size appears larger at the horizon. Lloyd Kaufman and Irvin Rock's 1962 experiments found that the illusion depends strongly on distance cues from the terrain; removing them makes the horizon Moon look smaller. Critics note that most people report the horizon Moon looks closer, not farther, which sits awkwardly with the theory. Other accounts invoke relative size compared with surrounding objects. Ross and Plug concluded, "No single theory has emerged victorious." The moon illusion is an everyday reminder that visual space has a shape of its own: the flattened sky dome is precisely the kind of finite, curved structure that Luneburg's model tried to capture.

The Ames room

The American scientist Adelbert Ames Jr. designed distorted rooms that look rectangular when viewed through a peephole but are actually irregular, with one far corner much more distant than the other. According to the Wikipedia article, Ames filed a U.S. patent in 1940, granted in 1944, for "Demonstrating and testing visual space perception"; accounts of when he built the first room range from 1934 to 1946. A person standing in one far corner appears giant and a person in the other appears tiny. The peephole removes stereopsis and motion parallax, leaving only the misleading perspective cues. The room shows that the visual system's assumptions about the shape of the space (rooms are rectangular) can override knowledge that adults are roughly the same size.

Space and its content

In physics, according to general relativity, space and the matter in it influence one another, but only on astronomical scales, which is irrelevant to vision. The Wikipedia article on visual space observes that in visual space, by contrast, content matters at every scale. Identical physical separations can appear different depending on the features that mark them out, especially in depth. A gap marked by richly textured surfaces may look different from the same gap between two isolated points of light.

This dependence on content is one reason a single geometry does not fit visual space. A geometry describes relations among points independent of what occupies them. Visual space seems to be built from the things in it, layer upon layer. The article concludes that the everyday three-dimensional visual space is built from many perceptual and cognitive layers on top of physiological representations.

The neural basis of visual space

Local sign and the role of connections

In the mid-nineteenth century, Johannes Müller emphasized the importance of the specific connections of nerve fibers in determining what we perceive. Hermann Lotze proposed the principle of local sign: each point on the retina carries a distinctive signature that the mind uses to assign the stimulus a location. The question of how location is encoded, rather than just presence or color, became central.

Retinotopic maps

The Wikipedia article on retinotopy traces the discovery that the visual pathway preserves the layout of the retina. In the late nineteenth century Hermann Munk, working with dogs, and David Ferrier, working with monkeys, found that lesions to the occipital and parietal lobes caused blindness. Around 1900 Salomon Henschen linked lesion locations in humans to areas of lost vision. In 1909 Tatsuji Inouye described visual field maps from head injuries suffered by soldiers in the Russo-Japanese War; his work, published in German, was largely ignored. A few years later Gordon Holmes, studying soldiers of the First World War, independently reached similar conclusions.

The result is that neighboring points on the retina project to neighboring points in the primary visual cortex, forming a retinotopic map. Modern functional magnetic resonance imaging maps many such areas in the human brain, including V1, V2, V3, V4 and regions in the occipital and parietal lobes.

Parallel processing and the binding question

A retinotopic map is not visual space. The Wikipedia article on visual space emphasizes that visual signals are analyzed separately for brightness, color, orientation, depth and other attributes in different neural populations, and that no single neuron represents both a feature and its exact location in a way that would constitute a unified scene. How these separate analyses combine into the unified visual space we experience is, the article states, beyond current neurophysiology.

Place cells and cognitive maps

A different line of research concerns the brain's representation of where the animal itself is. In 1971 John O'Keefe and Jonathan Dostrovsky recorded neurons in the rat hippocampus that fired only when the animal was in a particular part of its environment. These place cells led to the theory, set out in O'Keefe and Lynn Nadel's The Hippocampus as a Cognitive Map (1978), that the hippocampus holds a map of the environment. Later work by Edvard and May-Britt Moser identified grid cells in the entorhinal cortex with hexagonal firing patterns. The 2014 Nobel Prize in Physiology or Medicine was awarded to O'Keefe and to the Mosers for these discoveries.

The Wikipedia article on place cells notes that visual cues, especially distant landmarks such as walls, are central to how place cells form and maintain their fields. A 2014 review by Hartley, Lever, Burgess and O'Keefe, "Space in the brain," in Philosophical Transactions of the Royal Society B, is cited in the article on visual space as a possible route to understanding how spatial experience arises, though the article cautions that the process remains unclear. Place cells represent allocentric space (where I am in the world), while visual space is egocentric (how the world is laid out from my viewpoint); linking the two is a major open problem.

Why visual space matters beyond philosophy

Vision science and ophthalmology

Understanding how depth and layout are perceived matters for diagnosing and treating binocular vision disorders. Stereoblindness, strabismus and amblyopia all affect the construction of visual space. Luneburg developed his model at the Dartmouth College Eye Institute, an institution devoted to the study of vision.

Visual art and perspective

Painters have long exploited depth cues. Linear perspective, aerial perspective, occlusion and shading are all pictorial devices for creating an impression of depth on a flat surface. The question of why a correctly drawn perspective picture can look distorted when viewed from the wrong position is a question about how visual space is constructed from cues.

Virtual reality and displays

Head-mounted displays and stereoscopic cinema present slightly different images to each eye, exactly as Wheatstone's stereoscope did. Because such displays present images on a screen at a fixed optical distance while disparity signals other depths, the cues that normally agree in natural viewing, such as accommodation and convergence, can be put in conflict. Knowing which cues the visual system uses and how it combines them is essential to designing displays that look right and are comfortable.

Philosophy of mind

Visual space remains a test case for theories of perception. Is the visual world a construction of the brain, as Helmholtz's unconscious inference suggests? Is it a direct pickup of information in the light? Is visual experience representational, and if so, what geometry does it represent? Carnap's separation of formal, physical and perceptual space continues to frame these debates.

Common misconceptions

"Visual space is just physical space as seen."

They are closely aligned for practical purposes but not identical. Careful experiments show systematic deviations, such as the Hering-Hillebrand deviation of the horopter, the flattened sky dome, and the failure of perceived parallels to match physical parallels.

"Luneburg proved that visual space is hyperbolic."

Luneburg proposed a hyperbolic model based on limited data and assumptions about convergence. Follow-up experiments mostly did not support it, and the field now treats the geometry of visual space as variable and context-dependent.

"We see depth mainly because we have two eyes."

Binocular cues matter, especially for fine depth at close range, but many monocular cues, such as perspective, occlusion, motion parallax and texture, also provide depth. A person viewing with one eye still has all of the monocular cues listed above.

"The Moon looks bigger at the horizon because the atmosphere magnifies it."

Its angular size is essentially unchanged and, if anything, slightly smaller at the horizon. The effect is perceptual.

"Retinotopic maps are the brain's picture of visual space."

They preserve the layout of the retinal image, but visual space involves depth, constancy across eye movements, and integration of many features that retinotopic maps alone do not capture. How a unified visual space arises remains unexplained.

"Molyneux's question was settled by philosophy."

It was argued about for three centuries and given an experimental answer only in 2011, when Held and colleagues found that newly sighted patients initially could not match felt shapes to seen ones but learned to do so within days.

Frequently asked questions

What is the difference between visual space and the visual field?

The visual field is the part of physical space imaged on the retina at a moment. Visual space is the perceptual space in which seen things are located and related to one another.

Is visual space Euclidean?

Not exactly. For practical navigation it behaves nearly Euclidean, but experiments reveal deviations that depend on observer, distance, context and task. No single non-Euclidean geometry has been confirmed.

Who first proposed that visual space is non-Euclidean?

Reid's 1764 geometry of visibles described visible figures as equivalent to figures on a sphere, which has a non-Euclidean geometry, though whether Reid intended a non-Euclidean geometry is debated. Mach argued in 1901 that physiological space differs from geometric space. Luneburg's 1947 model is the best-known explicit proposal of a specific non-Euclidean (hyperbolic) geometry.

How do experimenters study visual space?

By asking observers to make perceptual settings, such as making lines look parallel or points look equally spaced, and then measuring the physical arrangement that satisfies the judgment.

Which depth cues give absolute distance?

According to the Wikipedia article on depth perception, only convergence, accommodation and familiar size. The rest give relative depth.

What is the horopter?

The set of points in space whose images fall on corresponding points of the two retinas, so that they are seen in the same visual direction by both eyes. Points near the horopter, within Panum's fusional area, are seen as single.

Is visual space innate or learned?

Both views have long histories. Berkeley argued that distance perception is learned; Kant treated space as a form of intuition the mind brings to experience. Evidence such as the Held and colleagues study suggests that some cross-modal links require experience but are learned rapidly.

What is visual space agnosia?

A neurological condition in which a person has difficulty perceiving spatial relations among seen objects despite intact basic vision, showing that the construction of visual space is a specific brain function.

Summary

Visual space is the space of visual experience: the ordered arrangement of seen things as they appear to an observer. It is distinct from physical space and from the visual field, and Carnap's 1922 division of formal, physical and perceptual space provides a useful frame. Philosophers from Ibn al-Haytham to Berkeley, Reid and Kant debated how we see distance and what structure seen space has. Helmholtz's theory of unconscious inference and Fechner's psychophysics turned the question into an experimental one, and Luneburg's 1947 hyperbolic model offered a bold geometric answer that later data mostly did not support. Modern research shows visual space to be constructed from many depth cues, binocular and monocular, combined in ways that depend on context, producing a world that is accurate enough for action yet shows systematic distortions such as the moon illusion and the deviations of the horopter. Its neural basis begins with retinotopic maps and extends to the hippocampal place and grid cells, but how the brain assembles a unified visual space remains one of the open problems of neuroscience.

Sources

• Visual space (Wikipedia)
• Stanford Encyclopedia of Philosophy, Molyneux's Problem
• Stanford Encyclopedia of Philosophy, Thomas Reid
• Horopter (Wikipedia)
• Depth perception (Wikipedia)
• Stereopsis (Wikipedia)
• Moon illusion (Wikipedia)
• Emmert's law (Wikipedia)
• Ames room (Wikipedia)
• Unconscious inference (Wikipedia)
• Rudolf Luneburg (Wikipedia)
• Retinotopy (Wikipedia)
• Place cell (Wikipedia)