Visual Acuity

Abstract

Visual acuity is a measure of visual perception: the finest spatial detail the eye can resolve, conventionally the smallest gap or stroke width that can be told apart. Because a feature's difficulty scales with the angle it subtends at the eye, acuity is expressed as an angle — the minimum angle of resolution — and recorded in Snellen or logMAR notation on a graded letter chart. Its ceiling is set jointly by the eye's optics, by the spacing of the foveal cones that sample the image, and by neural processing, so that resolution acuity approaches the sampling limit of the cone mosaic while hyperacuity for relative position runs an order of magnitude finer. This article sets out the definition and measurement of acuity, its optical and retinal limits, hyperacuity, contrast sensitivity, and the development and clinical assessment of vision.

Keywords: visual acuity, minimum angle of resolution, logMAR, hyperacuity

The line on the wall chart that a person can just read, and the one below it that dissolves into a blur, mark a threshold every sighted person carries: the limit beyond which the eye can no longer separate one detail from the next. Visual acuity is the psychophysical measure of that limit — the acuteness of spatial vision, quantified as the angular size of the smallest resolvable feature. It is at once the most routinely measured quantity in all of vision, printed on every optometric record as a fraction like 20/20, and a deep window onto how the optics of the eye, the mosaic of photoreceptors, and the cortex that reads them together set the resolution of sight (Elliott, 2016).

What Visual Acuity Is

Visual acuity is the capacity of the visual system to resolve fine spatial detail: to register that two nearby features are two, not one, or that a small target has a particular shape. Its defining move is to measure that capacity not in millimetres on the page but in the angle the critical feature subtends at the eye, because the same object casts a retinal image half as wide at twice the distance. A detail that spans one arcminute — one-sixtieth of a degree — at the eye projects the same retinal image whether it is a fine line viewed near or a broad stripe viewed far, so angular measure makes acuity a property of the observer rather than of the chart (Watson & Ahumada, 2012). The standard summary statistic is the minimum angle of resolution, the MAR: the angular size of the smallest feature the observer can reliably identify, quoted in arcminutes.

Acuity so defined comes in several operationally distinct kinds. Detection acuity asks whether a target is present at all; resolution acuity, the sense probed by a grating or a Landolt ring, asks whether two elements can be separated; recognition acuity, the sense the letter chart measures, asks which of several forms was shown. These do not all yield the same number, because identifying a letter requires more than merely resolving its strokes, and a model of acuity must specify the task as well as the stimulus (Watson & Ahumada, 2012). Figure 1 shows the geometry that underlies the classic 20/20 standard: a letter built on a five-by-five grid whose whole height subtends five arcminutes, so that each stroke and gap subtends one.

Types of Visual Acuity

MeSH files two narrower descriptors directly beneath Visual Acuity, shown in Table 1. They are a heterogeneous pair, and the grouping is a reminder that the tree is an indexing classification for the literature, not a theory of how vision is organised: the two subtypes are not mutually exclusive alternatives, nor are they both kinds of acuity in the strict sense. Contrast sensitivity extends the notion of resolution from the single high-contrast limit of the letter chart to the full range of contrasts and spatial scales, while emmetropia is a refractive state of the eye that MeSH places here as the optical precondition for normal acuity rather than a variety of acuity as such. Link each to its own article where one exists; otherwise the term is listed as the descriptor names it.

Measuring Acuity

The instrument everyone recognises is the Snellen chart, and the notation it produced still dominates the clinic. A Snellen acuity is written as a fraction whose numerator is the testing distance and whose denominator is the distance at which the smallest legible line would subtend the reference angle: 20/20 (or 6/6 in metres) means the observer reads at twenty feet what a standard eye reads at twenty feet, and 20/40 means they need to be at twenty feet to read what the standard eye reads at forty. The fraction is thus an inverse angular measure — the larger the denominator, the coarser the detail resolved. Converting it to the minimum angle of resolution makes the geometry explicit: MAR in arcminutes equals the denominator divided by the numerator, so 20/20 is one arcminute and 20/200 is ten.

The Snellen chart is nonetheless a poor measuring instrument, and much of modern acuity assessment is a reaction against its defects. Its lines carry different numbers of letters, the spacing between letters and lines is irregular, and the size steps are unequal, so a change of one line means different things in different parts of the chart (Elliott, 2016). Bailey and Lovie set out the design principles that fixed this: five letters on every line, inter-letter and inter-line spacing proportional to letter size, and a constant logarithmic size ratio of about 0.1 log unit between lines, so that every line is the same task differing only in scale (Bailey & Lovie, 1976). Acuity on such a chart is reported as the base-ten logarithm of the MAR — logMAR — on which 20/20 is 0.00, each line is 0.10, and lower or negative scores mean sharper vision. Because every letter carries equal weight, acuity can be scored letter by letter rather than line by line, which is more reliable and more sensitive to small change. The standardised version of this chart adopted for clinical trials, the ETDRS chart, made the logMAR design the reference method for research (Ferris, Kassoff, Bresnick, & Bailey, 1982). The first demonstration lets the reader simulate an acuity chart, degrade the eye's resolution, and read off the resulting Snellen and logMAR scores.

Optical and Retinal Limits

Two stages set the ceiling on resolution acuity before the cortex is even reached: the quality of the retinal image the eye's optics can form, and the fineness with which the photoreceptor mosaic samples that image. The optics come first. Even a well-corrected eye is not diffraction-limited at a large pupil, because higher-order aberrations — departures from a perfect lens that spectacles cannot correct — blur the retinal image, and their measurement showed that the eye's optical quality falls short of the retina's sampling capacity for pupils above a few millimetres (Liang & Williams, 1997). At the pupil sizes typical of bright viewing, though, the optics deliver an image fine enough that the limit passes to the retina.

There the grain of the cone mosaic takes over. The foveal cones form a dense, roughly triangular lattice, and the classic anatomical count put their peak density near the centre of the fovea at close to two hundred thousand per square millimetre, corresponding to a centre-to-centre spacing of about half an arcminute (Curcio, Sloan, Kalina, & Hendrickson, 1990). A regular sampling array can faithfully represent detail only up to its Nyquist limit — a spatial frequency of roughly sixty cycles per degree for that spacing — and detail finer than this is not merely lost but misrepresented as a spurious coarser pattern, an artefact called aliasing. By bypassing the eye's optics with laser interference fringes projected directly on the retina, researchers drove the input past the Nyquist limit and observers duly reported the tell-tale aliased patterns, direct evidence that foveal sampling, not the optics, sets the resolution ceiling (Williams, 1985). Imaging the living mosaic later closed the loop: with adaptive optics correcting the aberrations, an observer's finest resolvable grating matched the Nyquist limit of that observer's own measured cone spacing (Rossi & Roorda, 2010). The mosaic is not fixed across eyes, however; foveal cone density varies substantially between people and declines with the axial elongation of the eye that accompanies myopia, so the sampling limit itself is an individual quantity (Wang et al., 2019). The second demonstration builds a patch of the cone lattice, lays a grating over it, and lets the reader raise the spatial frequency past the Nyquist limit to watch resolution give way to aliasing.

Hyperacuity

If the cone spacing sets the resolution limit, a puzzle follows at once: on certain tasks people are reliably far more precise than that spacing should allow. Asked whether the upper and lower halves of a broken vertical line are offset, an observer can detect a misalignment of only a few arcseconds — a fraction the width of a single foveal cone, and roughly ten times finer than the resolution acuity of the same eye. Westheimer named this class of judgements hyperacuity and showed that its precision is optimal when the features to be compared lie a few arcminutes apart, falling off as they are brought together or pulled far apart (Westheimer & McKee, 1977). Vernier acuity, the offset judgement, is the canonical example; stereoacuity, the detection of small binocular depth differences, belongs to the same family.

The resolution of the puzzle is that hyperacuity is not resolution at all. Nothing about it requires two features to be separated on the retina; it requires the visual system to estimate and compare the positions of features that are already resolved. Because the blurred image each feature casts covers many cones, the pooled response of that population encodes the feature's centroid with a precision much finer than the spacing of the individual samplers, and the nervous system reads out this population-coded position. Hyperacuity is therefore a neural computation layered on top of the optical and retinal machinery, and its exquisite precision is a property of the cortex's use of the signal rather than of the retina's grain. That distinction matters clinically, because tasks of position and pattern degrade in ways plain letter acuity can miss. The third demonstration presents a broken line whose two segments the reader can offset by a controllable amount, reading the threshold in arcseconds against the width of a foveal cone.

Contrast Sensitivity

The letter chart measures resolution at a single, very high contrast: black ink on white paper. But natural scenes are full of detail at every contrast and every scale, and a fuller description of spatial vision asks how much contrast is needed to see a pattern of a given coarseness. Campbell and Robson made this quantitative by treating the visual system as a set of channels tuned to spatial frequency and probing it with sinusoidal gratings, plotting the reciprocal of the threshold contrast against spatial frequency to yield the contrast sensitivity function (Campbell & Robson, 1968). That function is band-pass: sensitivity peaks at a few cycles per degree, falls for coarser patterns, and falls to zero at a high-frequency cut-off. That cut-off — the finest grating visible at maximum contrast — is the acuity limit, so ordinary acuity is a single point on the contrast sensitivity function, its rightmost intercept.

Reading acuity as one point on a curve explains why two eyes with identical letter acuity can see the world very differently. A person can retain a normal high-frequency cut-off, and so a normal Snellen acuity, yet have depressed sensitivity in the middle of the range, losing the low-contrast detail that letter charts never test; comparing the shape of the function across normal and low vision shows that the loss need not be a simple horizontal shift (Chung & Legge, 2016). The contrast sensitivity function thus generalises acuity, embedding the single resolution threshold in a fuller map of spatial vision and revealing deficits that a chart of high-contrast letters leaves invisible.

Development and Clinical Assessment

Acuity is not fixed at birth; it is built. Newborn acuity is poor — on the order of one cycle per degree, some thirty times coarser than the adult limit — and it improves rapidly over the first months and years as the fovea matures and the cones pack more densely. Measuring it in a preverbal infant demanded a method that needs no report, and forced-choice preferential looking supplied one: an infant shown a grating beside a blank field of matched mean luminance will look preferentially at the grating while it is resolvable, and an observer who cannot see the display judges from the infant's gaze where the pattern was, yielding a threshold (Teller, 1979). The same logic, refined into acuity cards and other calibrated tools, underlies contemporary paediatric acuity testing, though the choice of test materially affects the number obtained and normative values are age-specific (Anstice & Thompson, 2014).

The developmental window that builds acuity is also where it can fail. Amblyopia — a reduction in acuity, uncorrectable by lenses, that follows abnormal visual experience early in life, such as a squint or unequal refractive error between the eyes — is the commonest developmental disorder of spatial vision, and its deficit extends beyond letter acuity to positional and contrast tasks, reflecting disordered cortical processing rather than a simple optical fault (Levi, 2020). Part of that wider deficit is crowding: a letter that is easily identified in isolation becomes far harder to recognise when it is flanked by neighbours, a limit on object recognition distinct from resolution that is present in normal peripheral vision and markedly exaggerated in amblyopia, and that operates on the very letter charts whose lines pack optotypes side by side (Whitney & Levi, 2011). Clinically, the reach of acuity testing has widened as the tools have become portable: a validated smartphone-based test can measure acuity to a standard comparable with a chart, extending reliable assessment to community and low-resource settings where a calibrated chart and lane are unavailable (Bastawrous et al., 2015). Across the lifespan, then, the single number on the chart rests on a developmental trajectory, a vulnerable critical period, and an expanding set of methods for reading it out.

Worked Example

The notation becomes concrete when the conversions are worked by hand, and the acuity-chart demonstration reproduces the arithmetic. Start from a Snellen acuity of 20/40. The minimum angle of resolution is the denominator over the numerator, 40 / 20 = 2 arcminutes, so the smallest resolvable stroke subtends twice the 20/20 reference. Its logMAR value is the base-ten logarithm of that MAR, log10(2) = 0.30; the two lines by which 20/40 sits below 20/20 on a logMAR chart are exactly the 0.10-per-line steps that Bailey and Lovie built in. A 20/20 acuity gives MAR = 20 / 20 = 1 arcminute and logMAR = log10(1) = 0.00, the origin of the scale, while 20/200 gives MAR = 10 arcminutes and logMAR = 1.00.

Now connect the number to the retina. One arcminute of visual angle corresponds on the retina to about 4.8 micrometres — roughly the diameter of a single foveal cone. At 20/20 the critical stroke of a letter subtends one arcminute, so it falls across about one cone; resolving the gap in the reference grating means the mosaic must sample a pattern whose period is near two cones, which is the Nyquist limit again from the psychophysical side. A grating of sixty cycles per degree has a period of one arcminute, half of which — the light-to-dark half-cycle — is the same half-arcminute as the cone spacing, so 20/20 letter acuity and the sixty-cycle grating cut-off and the half-arcminute cone spacing are three descriptions of one limit. The hyperacuity threshold makes the contrast vivid: a Vernier offset of five arcseconds is one-twelfth of an arcminute, about 0.4 micrometres at the retina — far less than a cone's width — which no account based on resolving the offset on the mosaic could deliver, and which only a positional computation pooling across many cones can.

Discussion

Visual acuity is deceptively simple to measure and surprisingly deep to explain. As a number it is an angle, the minimum resolvable feature size, and its measurement is a solved problem once the chart is designed to sample that angle in equal logarithmic steps (Bailey & Lovie, 1976; Elliott, 2016). As a phenomenon it is the visible outcome of a chain of limits: an optical stage whose aberrations blur the image, a retinal stage whose cone mosaic samples it and cannot represent detail beyond its Nyquist frequency, and a cortical stage that reads the sampled signal (Liang & Williams, 1997; Curcio et al., 1990; Williams, 1985; Rossi & Roorda, 2010). That resolution acuity sits so close to the sampling limit of the mosaic is one of the tidy optimalities of the visual system: the optics are matched to the receptors, and the receptors to the task.

Yet the single chart number understates what spatial vision does. Hyperacuity shows that the cortex can localise features an order of magnitude more finely than the mosaic can resolve them, because position is computed from a population rather than read from a pixel (Westheimer & McKee, 1977). The contrast sensitivity function shows that resolution is only the high-frequency edge of a broader capacity, and that clinically important loss can hide in the mid-range where the letter chart is silent (Campbell & Robson, 1968; Chung & Legge, 2016). And the developmental and clinical picture shows that acuity is constructed over a critical period, disturbed in amblyopia, and now measurable far beyond the consulting room (Teller, 1979; Levi, 2020; Bastawrous et al., 2015). Acuity is, in the end, both the humblest of clinical measurements and a precise probe of where in the visual pathway the resolution of sight is set.

Common Misconceptions

20/20 vision means perfect or ideal eyesight.

20/20 is a reference standard, not a ceiling. Many healthy young eyes resolve 20/15 or better, and 20/20 letter acuity says nothing about contrast sensitivity, colour, depth, or peripheral vision, so it can coexist with real visual complaints (Elliott, 2016; Chung & Legge, 2016).

Acuity is limited only by the sharpness of the eye's optics.

At normal pupil sizes the ceiling is set by the spacing of the foveal cones, not the optics: detail finer than the mosaic's Nyquist limit is sampled as aliasing, and with the optics bypassed or corrected, resolution matches the cone spacing itself (Williams, 1985; Rossi & Roorda, 2010).

The eye can never see detail finer than a single photoreceptor.

For resolution that is broadly true, but hyperacuity tasks such as Vernier alignment reach thresholds far below the width of one cone, because they compute the relative position of features pooled across many receptors rather than resolving them on the mosaic (Westheimer & McKee, 1977).

Glossary

Aliasing.

The misrepresentation of spatial detail finer than a sampling array's Nyquist limit as a spurious coarser pattern; in the eye it appears when detail exceeds the resolving capacity of the cone mosaic.

Amblyopia.

A developmental reduction of acuity in one or both eyes, uncorrectable by lenses, arising from abnormal visual experience such as strabismus or unequal refraction during the critical period.

Cone mosaic.

The dense, roughly triangular lattice of cone photoreceptors that samples the retinal image; its centre-to-centre spacing in the fovea sets the Nyquist limit on resolution acuity.

Contrast sensitivity function.

The reciprocal of the threshold contrast plotted against spatial frequency; a band-pass curve whose high-frequency cut-off is the acuity limit.

Crowding.

The impairment in identifying an object, such as a letter, when it is surrounded by nearby flankers; a limit on recognition distinct from resolution acuity, weak at the fovea but strong in peripheral vision and in amblyopia.

Emmetropia.

The refractive state of an eye that focuses distant light exactly on the retina with accommodation relaxed, so that no corrective lens is needed for clear distance vision.

ETDRS chart.

The standardised logMAR letter chart of the Early Treatment Diabetic Retinopathy Study, with five equally legible letters per line and constant logarithmic size steps; the reference method for acuity in clinical trials.

Hyperacuity.

A class of spatial judgements, such as Vernier alignment, in which thresholds are finer than the cone spacing because the task computes relative position from a population of receptors rather than resolving features on the mosaic.

logMAR.

The base-ten logarithm of the minimum angle of resolution; the acuity scale of the Bailey–Lovie chart, on which 20/20 is 0.00 and each line is 0.10.

Minimum angle of resolution (MAR).

The angular size of the smallest feature an observer can reliably resolve, quoted in arcminutes; one arcminute corresponds to 20/20 acuity.

Nyquist limit.

The highest spatial frequency a sampling array can represent without aliasing, set by the sample spacing; for the foveal cones it corresponds to roughly sixty cycles per degree.

Preferential looking.

A forced-choice method for measuring acuity in preverbal infants, exploiting their tendency to gaze at a resolvable grating rather than a blank field of matched mean luminance to yield a threshold without a verbal report.

Snellen fraction.

An acuity notation whose numerator is the testing distance and denominator the distance at which the smallest legible line subtends the reference angle; 20/40 means twice the reference MAR.

Vernier acuity.

The precision with which the alignment of two abutting line segments can be judged; a hyperacuity, with thresholds of a few arcseconds.

Visual angle.

The angle a feature subtends at the eye, equal to its size divided by its distance; the unit in which acuity is expressed so that it is independent of viewing distance.

Key Researchers

Christine A. Curcio. University of Alabama at Birmingham (Department of Ophthalmology and Visual Sciences); she produced the landmark quantitative map of human photoreceptor topography, fixing the peak foveal cone density that underlies the spatial-resolution limit of vision. Faculty Page - Google Scholar - ORCID

David B. Elliott. University of Bradford (Bradford School of Optometry and Vision Science); a leading voice in clinical acuity measurement, he has argued for the logMAR chart over the Snellen chart and clarified the metrology of best-corrected visual acuity and letter-by-letter scoring in practice and trials. Google Scholar - ORCID

Dennis M. Levi. University of California, Berkeley (School of Optometry & Vision Science); he mapped the neural mechanisms of amblyopia, crowding, and Vernier acuity, showing how abnormal early visual experience degrades acuity and positional precision and how perceptual learning can recover some function in adulthood. Faculty Page - Google Scholar - Wikipedia - Wikidata - ORCID

Austin Roorda. University of California, Berkeley (Herbert Wertheim School of Optometry & Vision Science); he developed adaptive optics scanning laser ophthalmoscopy for cellular-resolution imaging of the living cone mosaic and showed directly that foveal cone spacing sets the limit of human resolution acuity. Faculty Page - Google Scholar - ORCID

Gerald Westheimer. University of California, Berkeley (Professor Emeritus, Molecular and Cell Biology); he coined the term hyperacuity and established that observers localise spatial offsets an order of magnitude finer than the smallest foveal cone spacing, showing that fine spatial vision is a neural computation of relative position rather than a limit set by the retinal mosaic. Faculty Page - Wikipedia - Wikidata - ORCID

David R. Williams. University of Rochester (Center for Visual Science; Institute of Optics); he pioneered adaptive optics for the human eye, enabling in vivo imaging of individual cone photoreceptors, and used laser interference fringes to reveal aliasing in human foveal vision, mapping the sampling limits that set the ceiling on visual acuity. Faculty Page - ORCID

Frequently Asked Questions

What does 20/20 vision mean?
It means the eye resolves at twenty feet the detail a standard eye resolves at twenty feet, a critical feature subtending one arcminute of visual angle. It is a reference standard, not a maximum: many healthy eyes see 20/15 or sharper, and 20/20 letter acuity leaves contrast, colour, and depth untested (Elliott, 2016).

What is the minimum angle of resolution?
The minimum angle of resolution, or MAR, is the angular size of the smallest feature an observer can reliably resolve, measured in arcminutes. It equals the Snellen denominator divided by the numerator, so 20/20 is one arcminute and 20/200 is ten; its base-ten logarithm is the logMAR score (Bailey & Lovie, 1976).

Why is the logMAR chart better than the Snellen chart?
The Snellen chart has unequal numbers of letters per line, irregular spacing, and unequal size steps, so a one-line change means different things across the chart. The logMAR (Bailey–Lovie) chart uses five letters per line with proportional spacing and constant logarithmic steps, allowing reliable letter-by-letter scoring (Bailey & Lovie, 1976; Ferris et al., 1982).

What sets the physical limit on visual acuity?
At ordinary pupil sizes the limit is the spacing of the foveal cones, which sample the retinal image. The finest resolvable grating sits near the mosaic's Nyquist limit of about sixty cycles per degree; finer detail is aliased, and with the optics corrected, resolution matches the measured cone spacing (Curcio et al., 1990; Williams, 1985; Rossi & Roorda, 2010).

What is hyperacuity?
Hyperacuity is a class of spatial judgements, such as Vernier offset and stereoscopic depth, with thresholds of a few arcseconds, far finer than the cone spacing. It is possible because the task computes the relative position of already-resolved features from a population of receptors, rather than resolving them on the mosaic (Westheimer & McKee, 1977).

How is contrast sensitivity related to acuity?
Acuity is a single point on the contrast sensitivity function: its high-frequency cut-off, the finest grating visible at maximum contrast. The full function shows sensitivity across all spatial scales, so an eye with normal acuity can still have depressed mid-range sensitivity that the letter chart never reveals (Campbell & Robson, 1968; Chung & Legge, 2016).

How is visual acuity measured in babies?
By preferential looking: an infant shown a grating beside a blank field of matched luminance looks toward the grating while it is resolvable, and an observer blind to the display infers the threshold from the infant's gaze. Newborn acuity is around one cycle per degree and improves rapidly over the first years (Teller, 1979; Anstice & Thompson, 2014).

What is amblyopia?
Amblyopia is a developmental reduction in acuity, uncorrectable by lenses, that follows abnormal visual experience early in life such as a squint or unequal refraction between the eyes. Its deficit reflects disordered cortical processing and extends beyond letter acuity to positional and contrast tasks (Levi, 2020).

References