No, babies don’t see in black and white — and their focus isn’t fixed at 25 cm

Two claims have circulated for decades in parenting magazines, pregnancy-tracking apps and the “what your baby sees” images shared by the million. The first says a newborn perceives the world in shades of grey. The second, even more appealing, says they only see clearly at twenty-five centimetres, exactly the distance to their mother’s face during feeding. Both are false, and this has been known for a long time: infant vision has been measured in laboratories since the 1970s, without the baby needing to say a word. What these measurements show is stranger, and far more useful to know, than the legend.

WHERE THE MYTH COMES FROM

Why everyone believes babies see in grey

The idea has a respectable origin: it comes from a simplification of real findings. In the 1970s, the first serious measurements of infant vision showed very low acuity and a sensitivity to colour that was hard to demonstrate. A shortcut took hold — no measurable colours, therefore no colours — and it was then passed on from leaflet to leaflet, then from website to website, long after the protocols had been refined.

The second driver is visual. For around fifteen years, series of images have circulated claiming to reproduce a baby’s perception month by month: a blurred, desaturated photo of a face, then a little less blurred, then sharp. These composites are shared widely because they are instantly understandable. The problem isn’t that they exaggerate, it’s that they conflate two very different things: the resolution of an image and the experience of an observer. No one can reproduce what an infant perceives, for the simple reason that their perception can’t be compared to anything we know — they aren’t simply missing a few pixels from an image that is otherwise identical to ours.

What can be measured, however, can be measured very well. And the picture that emerges is of a visual system that works from birth, that already sorts what it looks at, but each of whose components — the optics, the retina, the cortex — follows its own maturation timetable.

THE METHOD

How to measure the sight of someone who can’t talk

The whole difficulty comes down to one sentence: visual acuity is normally measured by asking someone to name letters. An infant names nothing. The solution that was found is remarkably elegant, and it rests on a spontaneous behaviour: placed in front of two surfaces, a baby preferentially looks at the one containing a pattern rather than the one that’s plain.

The child is shown a card with one half in black-and-white stripes and the other in a grey of the same average brightness. If the child turns their eyes towards the stripes, it means they can see them. The stripes are narrowed, the test repeated, and the examiner looks for the finest stripe width beyond which the two halves look identical to the child. This limit, expressed in cycles per degree — the number of pairs of bars the eye can still separate within one degree of visual angle — is their acuity. The method is standardised as sets of cards, and detailed norms have been published for every age band between one month and four years.

The order of magnitude, first. In a cohort of children from southern China tested with Teller Acuity Cards (TAC II), average binocular acuity reaches 3.7 cycles per degree at twelve months, 7.4 at two years and 12.0 at three years; measured monocularly, the same children score 2.9, 6.9 and 10.8. An adult can distinguish around thirty. In other words, a one-year-old sees detail roughly eight times less finely than they will as an adult, and it will still take years to close the gap. This is a very long way from blindness, and just as far from fully developed vision.

These figures deserve a caveat: they vary from one population to another and depending on which version of the cards is used. The historical norms established in the United States on 460 children don’t necessarily map onto them, which is a reminder that an acuity norm is a local clinical tool, not a universal biological constant.

FIRST MYTH

Red, yes. Blue, almost never.

It’s an experiment that settles the black-and-white question, and it’s disarmingly simple. Newborns are shown a coloured patch on a grey background, and the brightness of the grey is systematically varied. This precaution is decisive: a baby could tell a red patch apart simply because it’s darker than the background, without perceiving any colour at all. By sweeping through every possible brightness, this cue is eliminated — if the baby keeps responding, it really is the hue being detected.

The result, across thirty-six newborns and thirty-four one-month-olds: 74% of newborns succeed with a red patch. With green, only 36% do; with yellow, 25%; with blue, 14%. At one month, performance improves slightly, but red remains the only clearly established colour. The authors’ conclusion is measured: colour vision in the first weeks is very limited. It isn’t absent.

The difference isn’t trivial. “The baby sees in black and white” suggests an overall deficit that would correct itself all at once. The reality is a deficit that varies from one hue to another: red stands out from birth, blue much later, and hue discrimination becomes established over the first three months. The researchers who measured it attribute this less to a delay in one particular colour channel than to an overall immaturity, that of the retina’s cones as well as that of the circuits that compare their signals: the whole system gains in sensitivity, and the hues that are hardest to tell apart are the last to cross the threshold.

In practical terms, this explains an observation every parent makes: in the first few weeks, a mobile with sharp contrasts — black on white, red on white — captures attention better than a toy with soft, similar colours. It isn’t a matter of taste, it’s a matter of what clears the detection threshold.

SECOND MYTH

Focus isn’t fixed at twenty-five centimetres

This is the most tenacious legend, and the prettiest: the newborn supposedly sees clearly at one single distance, the one separating their face from that of the parent holding them. The story is appealing because it describes a well-designed nature. It doesn’t survive measurement.

The technique used is called photorefraction: a camera placed one or two metres away captures the light reflex coming back from the eye, and the shape of that reflex shows in real time where the eye is focusing. No contact, no cooperation required. In the key study, twenty-nine infants aged two to ten months were tested with targets presented in succession at 25, 33, 50 and 100 cm. Every single one adjusted their accommodation in the right direction at each change of distance. The speeds measured reached at least 4.6 dioptres per second, comparable to an adult’s. Before two months, the picture is less clear-cut: in another study, carried out at six weeks and at three months, some babies were already adjusting their focus when the target changed distance, while others kept a roughly fixed setting.

The mechanism behind this focusing is the same as in adults: the lens changes shape under the action of the ciliary muscle, and this variation is measured in dioptres. What isn’t mature in the first few weeks is the precision of the adjustment, and above all the fineness of the image the retina is able to make use of once focus is achieved. The infant focuses correctly on an object one metre away — it’s simply that, at one metre, they can’t see much, not because the image is poorly focused but because their resolution is low.

So where does the figure of twenty-five centimetres come from? Probably from an accurate observation that has been misinterpreted: that’s roughly the distance at which a face occupies enough of the visual field for its bold contrasts to be detectable by a low-resolution system. The baby doesn’t see better at that distance because their optics are somehow tuned to it; they see better there because the object is bigger.

INTERACTIVE MODULE

Five years for the world to come into focus

The module below brings this timeline together. Drag the slider: at each age, it shows visual sharpness expressed as a fraction of an adult’s, the state of colour vision, the state of depth perception, and what the child does at that stage. Values marked “measured” come from published norms; the others indicate a trend. The image on the left is an analogy, not a reconstruction.

THE SWITCH

Depth perception arrives all at once, between ten and nineteen weeks

Of all the milestones in visual development, this one is the most spectacular, because it doesn’t look like a gradual progression. Seeing in depth requires the brain to compare the two slightly offset images sent by the two eyes and extract depth information from them. As long as the circuits that perform this computation aren’t working, depth information is simply missing; what the measurements make it possible to date is the moment they come into operation.

To date this, one team used random-dot stereograms — images that contain no recognisable shape, except for a brain able to fuse the two eyes’ views, in which case a figure emerges from the noise. By recording the potentials evoked by these images in infants, the authors observed that the stereoscopic response appears between the tenth and nineteenth week. Yet the same babies had already been producing normal responses to ordinary checkerboard patterns for weeks. The eyes were therefore already seeing, each on its own side; what was missing was pooling the two together.

This switch visibly changes behaviour. It’s around this period that the child stops flailing at the air and starts to grasp what they’re aiming at: they finally know how far away the object is. This is also why an obstacle to binocular vision during this window — constant strabismus, a large difference in correction between the two eyes — has lasting consequences for depth perception, well beyond the period in which it’s present.

FROM THE FIRST HOUR

What a newborn chooses to look at

If a newborn sees so few details, you’d expect them to look at anything at random. It’s the opposite. A series of experiments carried out in maternity wards showed that a baby less than an hour old tracks a face-like pattern with their eyes and head — an oval bearing three patches arranged like two eyes and a mouth — further than they track the same oval with the patches randomly rearranged, or a blank oval. The difference is modest but reproducible, and it was confirmed in a second maternity unit by the same team.

The most intriguing detail comes from the rest of the study. Following the same children for five months, the authors observed that this preference declines during the second month. It doesn’t strengthen, it disappears — replaced by an interest in faces of an entirely different nature, a learned one this time, which distinguishes individuals and reads expressions. Everything happens as if a rudimentary, hard-wired mechanism served as a bootstrap, steering the newborn’s gaze towards faces while the system learns, then fading out once its job is done.

This is the best answer to the question, “but does my baby recognise me, then?” In the first few days, they don’t recognise a face in the sense we mean by that; but they are equipped to look for one, which is no small thing.

THE EYE ITSELF

Nothing is finished at birth, not even the optics

It’s tempting to picture an eye that’s complete at birth, with only the brain left to learn how to use it. Histology says otherwise. The fovea — the tiny dip at the centre of the retina that carries all fine vision — begins forming around the twenty-fifth week of pregnancy, keeps widening after birth, and doesn’t take on its adult shape until around fifteen months. The central cones, for their part, continue maturing for much longer: their segments, which measure the same as those in the periphery at fifteen months, are four times longer by age thirteen.

The optics change just as much. A longitudinal study following children from three months to six-and-a-half years shows that eye length increases by 3.35 mm on average over that period, while the power of the lens decreases and that of the cornea stabilises after an initial fall. These three changes aren’t independent: they compensate for one another. This is the adjustment, called emmetropisation, that brings an eye which is generally too short at birth — and therefore long-sighted — towards the correct focal length. Most of it takes place within the first two years.

When this adjustment later overshoots in the other direction, the eye becomes too long and the child becomes short-sighted. Significant uncorrected astigmatism during early childhood disrupts development in the same way, by durably depriving the retina of a sharp image in one particular orientation. This is what shows why the period is sensitive: it isn’t just that the child sees poorly at the time, it’s that the system builds itself on what it receives.

One final figure gives a sense of the real timescale. Contrast sensitivity — the ability to perceive faint shapes, not sharp stripes — gains roughly two logarithmic units between infancy and adulthood, and still hasn’t reached its adult level by age eight. A more recent analysis, covering close to two thousand measurements, still finds a doubling between four and eighteen, 90% of it achieved before age twelve; its authors note, however, that at least half of this gain, perhaps all of it, may reflect the child’s attention during the test rather than their vision. Most of it is played out in the first few years, but visual development is measured in years, not school terms.

WHAT THIS MEANS IN PRACTICE

Why this timeline is a medical clock

Everything above could remain a laboratory curiosity. It isn’t one, for a reason that paediatric cataract surgery has quantified with useful brutality.

Dense bilateral congenital cataract deprives the retina of a sharp image at the exact moment the visual cortex is wiring itself up. One study followed thirty-seven infants operated on before thirty-one weeks of age and measured their acuity at five years or more. The model that best describes the results includes a fourteen-week turning point: before it, every three-week delay in surgery costs, on average, one line of permanent acuity; after it, the outcome no longer depends on the timing of the operation and plateaus at around 20/80. A second, independent series finds the same relationship between the duration of deprivation and the visual outcome. The authors found no latency period during which one could wait without consequence.

This is the very definition of a critical period, a concept neurobiology has since dissected extensively: the plasticity of the visual cortex isn’t simply lost with age, it’s actively locked down by molecular mechanisms that get put in place over the course of development. Understanding these locks is today one of the most active avenues for treating amblyopia beyond childhood.

Concretely, this means there is a short list of situations in which one does not wait and see in an infant:

• A white or greyish reflection in the pupil, on a flash photo or in daylight, instead of the usual orange-red reflex. This is the sign the red-reflex test carried out in maternity wards is looking for, and it requires an examination without delay.
• Constant strabismus, or any strabismus that persists beyond the very first few months, or that always affects the same eye.
• An eye that looks distinctly bigger than the other, with watering and discomfort in light: this trio suggests congenital glaucoma.
• An absence of visual tracking when age would make it expected, or a child who doesn’t fix their gaze on faces.
• A drooping eyelid that covers the visual axis, whatever the cause.

Outside these situations, ordinary follow-up is enough — which never rules out seeing a doctor for a red, painful or injured eye, or strabismus that appears suddenly. The French recommendations for visual screening in children, published by a working group of paediatric ophthalmologists, combine systematic examinations at key ages with targeted referral in response to a warning sign or a family history. The principle is this: look early for objective signs, then measure eye by eye as soon as the child becomes testable, at around three years old.

THE TAKEAWAY

A system that sorts before it sees clearly

The real surprise in all this isn’t that the baby sees poorly. It’s the order in which the functions appear. Sorting comes before resolution: a newborn turns their gaze towards a face before being able to make out its features. Focusing is operational before the image is even worth focusing on. And depth perception, which seems the most sophisticated of the three, arrives earlier than visual sharpness, which will still take years to reach completion.

This order is not arbitrary. It matches what an organism needs to do in its first few months: spot a face, know where to reach, and learn. Sharpness, on the other hand, can wait — provided nothing blocks the channel while the wiring is being laid down. This is the whole reason early visual screening exists, and it’s also what makes the two legends from the start of this article more troubling than they seem: by repeatedly describing a baby who supposedly sees nothing, it starts to feel normal that they don’t react.

Frequently asked questions

Scientific sources

1. Systematic measurement of human neonatal color vision. Vision Res. 1994. PMID: 7941376
2. Development of chromatic discrimination in early infancy. Behav Brain Res. 1995. PMID: 7748507
3. Accommodation in infants as measured by photorefraction. Vision Res. 1987. PMID: 3447363
4. The development of accommodation. Vision Res. 1997. PMID: 9205713
5. Development of stereopsis and cortical binocularity in human infants: electrophysiological evidence. Science. 1981. PMID: 7268443
6. Newborns’ preferential tracking of face-like stimuli and its subsequent decline. Cognition. 1991. PMID: 1786670
7. Visual acuity in human infants: a review and comparison of behavioral and electrophysiological studies. Vision Res. 1978. PMID: 364823
8. Monocular acuity norms for the Teller Acuity Cards between ages one month and four years. Invest Ophthalmol Vis Sci. 1995. PMID: 7890497
9. Study to establish visual acuity norms with Teller Acuity Cards II for infants from southern China. Eye (Lond). 2021. PMID: 33235349
10. Histologic development of the human fovea from midgestation to maturity. Am J Ophthalmol. 2012. PMID: 22935600
11. Ocular Component Development during Infancy and Early Childhood. Optom Vis Sci. 2018. PMID: 30339640
12. Development of spatial contrast sensitivity from infancy to adulthood: psychophysical data. Optom Vis Sci. 1997. PMID: 9383792
13. Development of the spatial contrast sensitivity function (CSF) during childhood: analysis of previous findings and new psychophysical data. J Vis. 2020. PMID: 33275663
14. Development of human visual function. Vision Res. 2011. PMID: 21356229
15. The critical period for surgical treatment of dense congenital bilateral cataracts. J AAPOS. 2009. PMID: 19084444
16. Duration of form deprivation and visual outcome in infants with bilateral congenital cataracts. J AAPOS. 2010. PMID: 20227619
17. Critical periods in amblyopia. Vis Neurosci. 2018. PMID: 29905116
18. Evaluation of the red reflex: an overview for the pediatrician. World J Methodol. 2021. PMID: 34631483
19. Vision screening for children: Recommended practices from AFSOP. J Fr Ophtalmol. 2021. PMID: 33388188

Further reading

• Visual acuity: what the numbers really measure
• The retina: anatomy, examinations and pathologies
• Vision simulator: see what a corrected or uncorrected eye sees
• The eye assessment: what happens and what’s tested
• Dr Moïse Tourabaly — Ophthalmologist

Cachan practice · Tel. +33 1 45 47 08 11

Medical disclaimer

This article is intended for information purposes. A personalised ophthalmological opinion remains essential for any treatment decision.

The interactive module is a teaching schematic: the blur, contrast and saturation applied to the image are analogies intended to illustrate a point, in no way a reconstruction of an infant’s actual perception, which cannot be reproduced. The acuity values marked as measured come from the published norms referenced above and vary according to population and method; the others are orders of magnitude. This article does not replace a consultation, nor the systematic examinations of paediatric follow-up. The photographs are royalty-free illustrative images, unconnected to any real patient or to any particular condition. In the presence of a white or greyish pupillary reflex, constant strabismus, an eye that is enlarging with watering and photophobia, an eyelid covering the visual axis, an absence of visual tracking, or a child who doesn’t fix their gaze on faces, an ophthalmological examination is required without delay, without waiting for the next appointment. Dr Tourabaly sees children from three years of age, at the Cachan practice; the situations described here, which concern newborns and infants, call for a paediatric ophthalmology consultation.