Gibson, E. J., & Walk, R. D. (1960). The “visual cliff.” Scientific American, 202(4), 64-71.
The visual cliff experiment (Gibson & Walk, 1960) tested whether depth perception is innate. Researchers placed infants and newborn animals on a glass-topped table that created the illusion of a sudden drop, then recorded whether they would cross it.
Most infants and animals refused to cross it. This suggested that depth perception develops very early in life, rather than through learning.
Aim
Eleanor J. Gibson and Richard D. Walk (1960) investigated whether newborn animals and human infants could detect depth.
Gibson and Walk tested whether infants and young animals would crawl over an apparent drop. This would suggest depth perception is not present from birth.
If they refused instead, this would support a nativist view: that perceptual abilities are innate. Six months was the earliest age at which infants could be tested this way, since the procedure depended on independent crawling.
Key Takeaways
- The Apparatus: A safe glass-topped table creates the illusion of a sudden drop, letting researchers test depth perception without any real risk of falling.
- Human Findings: About 92% of 6- to 14-month-old infants refused to crawl onto the “deep” side even when their mother called them from across it.
- Animal Evidence: Newly hatched chicks, goat kids and lambs avoided the deep side completely from birth, supporting an innate explanation.
- The Crawling Confound: Because infants could only be tested from six months, once they could crawl, this confound (a factor besides the one studied) meant the study could not rule out earlier learning.
- Resolving the Confound: Later research measuring heart rate in two-month-olds (Campos, Langer & Krowitz, 1970) found they could already perceive the depth difference, long before they could crawl.
- Perception vs. Fear: Depth perception itself appears very early. Wariness of a drop builds up later, tracking an infant’s own movement experience (Dahl et al., 2013).
- A Fuller Picture: A historical review (Rodkey, 2015) found Gibson and Walk tested far more species than the popular chicks-goats-lambs story suggests, with mixed results.
Procedure
Gibson and Walk built a heavy glass-topped table, about four feet high, with a sturdy board running across its centre. A crawling infant or animal could be placed on this central board. There was no possibility of it actually falling.
One side, the “shallow side”, looked solid: a chequered pattern sat directly beneath the glass.
The other, the “deep side”, showed the same pattern several feet below on the floor. The glass itself stayed unbroken across both halves. This created the visual impression of a sudden drop.
Experiment 1
Sample: 36 infants ranging in age from six months to 14 months. Their mother also participated in the experiment.
The independent variable (IV) was whether the infant was called by its mother from the cliff side or the shallow side (of the visual cliff apparatus).
The dependent variable (DV) was whether or not the child would crawl to its mother.
This was a repeated measures design because the infant was called from both the cliff side and the shallow side of the apparatus.
Experiment 2
Using the same apparatus, Gibson and Walk tested chicks, lambs, and kids (young goats), all less than 24 hours old.
The studies using other species are quasi (laboratory) experiments. The naturally occurring independent variable (IV) was the animal species, e.g., rat/chick/lamb/kitten.
The dependent variable (DV) was whether the animal preferred the shallow side or the deep side of the visual cliff apparatus.
When animals were placed directly onto the deep side by hand, they typically froze into a defensive position rather than moving at all. This did not change with repeated trials. The animals never learned that the glass was actually safe to stand on.
Other species were also tested, including rats and kittens. Rats rely on touch from their whiskers rather than vision, and showed little preference for either side, even with their whiskers removed (Rodkey, 2015).
Kittens, tested several weeks after birth, showed a clear preference for the shallow side. Dark-reared kittens did not. But they matched others after about a week of light.
The rats were also tested with a uniform grey surface instead of the chequered pattern. This removed a key visual cue, to see whether the pattern itself was essential for perceiving depth.
Findings
- Human infants: Even when their mother called them, about 92% (33 of the 36) infants refused to cross onto the deep side. The few who did so typically backed onto it by accident rather than crawling forward deliberately.
- Chicks, lambs and kids: None crossed to the deep side. When placed there directly by hand, they froze into a defensive position, even after repeated trials.
- Rats: Rats use touch from their whiskers, not vision, to move around. They showed little preference for the shallow side even in darkness or with their whiskers removed (Rodkey, 2015).
- Kittens: Kittens tested at about four weeks showed a clear preference for the shallow side. Kittens reared in the dark until 27 days showed no preference at first, but matched light-reared kittens after about a week of light.
Conclusion
Gibson and Walk drew two main conclusions, one from the human data and one from testing animals.
The Human Evidence
Because the infants detected the danger on the “cliff” side, Gibson and Walk concluded that depth perception might be innate. It was, at the very least, present by the time an infant could crawl.
There was a catch, though. Gibson and Walk knew it. Human infants take several months to crawl, so they could in principle have learned to perceive depth during that time. The second experiment used animals to explore this possibility.
This apparent confound was not left unresolved for long.
Later research, described earlier, used heart-rate measures in infants too young to crawl and tested animals with no crawling age to speak of at all. Both approaches pointed the same way: toward depth perception being present very early in life.
Depth Perception Across Species
Animals can judge depth as soon as they become mobile, whether that is immediately after birth or hatching, or somewhat later. This ability does depend on some visual experience, meaning time spent in the light. Recovery from visual deprivation is quick, though, compared with how long the deprivation lasted.
Together, this suggests depth perception is innate.
Any species must develop depth discrimination by the time its members become independently mobile. The chick and the goat do this within a day. The rat and the cat take three to four weeks. The human infant takes six to ten months.
This fits a simple survival logic. An organism that perceives and avoids a drop before its first encounter with one is far more likely to survive. That logic applies whether the encounter comes on day one, as for a chick, or six months in, as for a human infant.
Later Research: Resolving the Learning Confound
Gibson and Walk’s human infants were already crawling, so their study could not rule out learning. Two later lines of research closed this gap by testing infants and behaviours that could not have been learned.
Testing Infants Too Young to Crawl (Campos, Langer & Krowitz, 1970; Campos et al., 1978)
Aim: Gibson and Walk’s method needed crawling infants. This meant it could not show whether depth perception, or fear of a drop, appears earlier. Campos, Langer and Krowitz (1970) solved this with heart rate, a response that needs no voluntary movement at all. A follow-up study came in 1978.
Method: In 1970, pre-locomotor infants were placed onto the shallow side, then the deep side, while electrodes recorded their heart rate throughout. The 1978 study used the same apparatus and measure. This time, though, the infants already had weeks or months of crawling experience.
Results: The pre-locomotor infants’ heart rate fell on the deep side in 1970, the pattern linked to interest rather than fear. Distress was rare.
The 1978 study found the opposite: infants with crawling experience showed a heart-rate rise on the deep side, the pattern linked to fear.
Conclusion: A heart-rate change in either direction shows an infant is registering the difference in depth. Even the youngest infants could tell the two sides apart.
But only the infants with crawling experience showed fear. Perceiving a drop and fearing it develop on different timetables.
Self-Produced Movement and Learning to Fear Heights (Dahl et al., 2013)
Aim: If fear of drops is learned, what do infants actually learn from? Dahl and colleagues focused on self-produced locomotion: moving one’s own body through space.
Method: Infants too young to crawl practiced steering powered go-carts by joystick for several weeks. A comparison group had no such practice. Researchers lowered both groups toward a real 1.3-metre drop-off while recording heart rate. The infants were tested separately in a “moving room” that simulated being pushed forward.
Results: Go-cart-experienced infants showed a clear heart-rate rise, about five beats per minute, toward the drop-off. They also recoiled from the moving room. Infants without practice showed neither response.
Conclusion: Learning to fear heights depends on practicing control of one’s own movement, whether by crawling, walking or driving a go-cart. It does not depend simply on getting older or falling a certain number of times.
Reading the Mother’s Face at the Brink (Sorce et al., 1985)
Aim: Not every drop is clearly safe or clearly dangerous. Sorce, Emde, Campos and Klinnert (1985) tested what one-year-olds do when the visual information alone does not settle the question, a phenomenon they called social referencing.
Method: One-year-olds were placed on a visual cliff apparatus modified so the drop looked ambiguous, neither obviously safe nor risky.
Their mother then posed one of several expressions, from joy to fear, as the infant looked up. The measure was whether the infant crossed to reach her.
Results: When the mother posed joy or interest, about three-quarters of infants crossed the ambiguous side to reach her. When she posed fear, almost none crossed, and many retreated instead. Anger and sadness held them back too.
Conclusion: By around a year old, infants actively seek a caregiver’s emotional signal to decide how to act at an ambiguous edge.
Avoiding a drop, then, is not a fixed reflex even once depth perception is in place. It can be overridden or confirmed by social information.
Critical Evaluation
Methodological Strengths
The procedure was a rigorously controlled laboratory test, offering a reliable and safe measure of depth perception. Gibson and Walk were able to eliminate or control the influence of other senses, such as touch from the rats’ whiskers, which helped ensure a valid test of visual perception.
Later studies built on this strength. The heart-rate and go-cart experiments added physiological and experimental measures to the original design. Together, these let researchers separate a single developmental achievement into its perceptual and motor parts.
Sample Size Limitations
The consistency of the results across a range of species, including humans, adds credibility to the findings. However, the human sample was small, and the six-to-14-month age range was wide enough that some infants had likely been crawling for months before they were tested.
There is a further limit worth noting. The strongest single-species evidence for innateness comes from precocial animals, such as chicks and goats, that can move within hours of birth.
Humans are altricial instead: helpless and immobile for months. A mechanism that is innate in a species built to move from birth may not be innate in a species built to be carried. It may not even mature on the same timetable.
Ethical Considerations
The apparatus’s key design feature is a continuous sheet of glass, so the “drop” is entirely illusory. This is what makes testing infants and newborn animals on an apparent cliff ethically defensible at all: no participant was ever at real risk of falling.
That said, a genuine welfare cost remained. Some infants in the original and later studies showed visible distress at the edge, including crying, refusing to move, and apparent fear.
Some animals were also deliberately placed on the deep side by hand and observed while frightened and immobile.
Although the mothers gave informed consent, these costs were real. They were mild and temporary. A modern ethical review would likely weigh them more explicitly than the original 1960s research culture did.
Contemporary Research
Since around 2015, this field has shifted away from treating the cliff as a test of one innate “fear module”. Adolph and Hoch (2019) reviewed decades of infant motor-development research instead.
They proposed that motor skill is embodied, embedded, enculturated and enabling: shaped by the body, the surroundings, caregiving practice, and each new skill it opens up.
The Perception-Action Reframing (Adolph & Hoch, 2019)
Applied to the cliff, this predicts something specific. An infant who has learned to avoid a drop while crawling will often fall over the identical drop soon after learning to walk.
The skill is not banked once and for all. It has to be relearned for each new posture.
The Forgotten Menagerie (Rodkey, 2015)
Historical work re-examining Gibson and Walk’s original records complicates the tidy chicks-goats-lambs textbook story. Rodkey (2015) traced the original experimental series and found the pair tested a far wider range of species, including rats, aquatic turtles, kittens, pigs, dogs and monkeys.
The results varied by species: aquatic turtles were the weakest avoiders, with about three-quarters choosing the shallow side, since life in water reduces the value of depth discrimination.
Locomotor Experience and Social Information (Karasik et al., 2016)
Aim: Karasik, Tamis-LeMonda and Adolph (2016) asked what determines how infants combine perceptual judgement with a caregiver’s social signal when deciding whether to descend a risky edge.
Specifically, they tested whether locomotor experience, rather than age alone, is the deciding factor. Age alone was not enough.
Method: Thirty infants were tested on an adjustable drop-off apparatus at their measured “borderline” height. They fell into three groups: 12-month crawlers, novice walkers, and 18-month experienced walkers. Their mother either encouraged or discouraged the descent.
Results: Experienced crawlers and walkers used their mother’s message selectively. They ignored it at clearly safe or clearly risky heights. But they were swayed by it at their own borderline height.
Novice walkers often attempted risky drop-offs regardless of her message and lost their balance more often.
Conclusion: Locomotor experience, not age, was what mattered. It determined how well an infant blended perceptual and social information about risk. Novice walkers are exposed to real risk precisely because they have not yet learned when to trust their own judgement and when to check in socially.
Perceptual Adaptation in Adults
The investigation of the nature-nurture issue in perception did not end with Gibson and Walk’s research. Their study only explored the plasticity of infant perception, leaving open the question of whether adult perception can also adapt.
This has, however, been investigated in several different ways. Stratton (1897) and Kohler (1962) used complex optical apparatus, such as inverting prisms, to change how their view of the world appeared.
In these situations, the world at first seems upside down or muddled, but the brain adapts over time and normal perception returns.
When the apparatus is removed, it takes time again. Perception slowly reverts to normal. Other, less dramatic changes can also be induced by shifting the field of view slightly to one side. Depth perception is then tested by, for example, how accurately a person can point to a target.
People adapt readily, usually within an hour. This only works, though, if they can actively interact with their environment. Participants who are instead pushed around in wheelchairs fail to learn to cope with the visual distortion (Held & Bossom, 1961).
Findings such as these suggest that, at least in some respects, depth perception is learned.
Real-World Applications
The visual cliff is not just a historic demonstration; its logic now informs infant safety guidance, clinical screening, and research methods across psychology.
Infant and Child Safety
Perceiving a drop is not the same as avoiding it. Wariness of heights lags behind both depth perception itself and each new motor skill (Campos et al., 1978; Dahl et al., 2013).
Risk peaks right after a locomotor transition. A newly mobile crawler, and especially a brand-new walker, is more likely to fall from a stair, a bed edge, or a changing table.
Experienced movers rarely fall this way.
This is why caregivers use stair gates and guard changing tables. A baby should also never be left unattended on a raised surface. A belief that “she’s always been careful” can stop being true the moment a child starts walking, since caution must be relearned in a new posture.
Risk peaks here too. A caregiver’s guard is often down exactly when a new walker looks capable enough to manage alone.
Clinical Screening for Preterm Infants
The paradigm has also been adapted for clinical use. Lin, Reilly and Mercer (2010) tested pre-walking infants born preterm alongside full-term infants of the same age on a modified visual cliff. They compared how each group approached the shallow side versus the deep side.
Both groups behaved in a clearly surface-dependent way. Each spent longer, and explored more, over the deep side.
However, the full-term infants used a wider range of motor strategies and avoidance behaviours than the preterm infants tested at the same age since birth.
This kind of comparison shows how a visual-cliff-style task can probe whether depth-guided caution emerges on the same timetable in atypical early development. Screening tools like this one help clinicians spot atypical development early, when support can do the most good.
A Durable Research Tool
This tool has uses well beyond nature versus nurture. Because it works across species with only minor changes, researchers have used it to study depth perception and visual development more broadly.
It has also served as a way to measure recovering visual function after an experimental change to the visual system.
The cliff is as much a method as a finding. It shows how to turn an invisible perceptual capacity into an observable behaviour, such as crossing, freezing, or a change in heart rate. It remains a standard teaching example for how psychologists weigh nativist versus empiricist claims about perception.
This flexibility explains a lot. The visual cliff is still a working method today, more than sixty years after Gibson and Walk built it. It remains far more than a museum piece.
References
Adolph, K. E., & Hoch, J. E. (2019). Motor development: Embodied, embedded, enculturated, and enabling. Annual Review of Psychology, 70, 141-164.
Campos, J. J., Langer, A., & Krowitz, A. (1970). Cardiac responses on the visual cliff in prelocomotor human infants. Science, 170 (3954), 196-197.
Campos, J. J., Hiatt, S., Ramsay, D., Henderson, C., & Svejda, M. (1978). The emergence of fear on the visual cliff. In M. Lewis & L. A. Rosenblum (Eds.), The development of affect (pp. 149-182). Springer US.
Dahl, A., Campos, J. J., Anderson, D. I., Uchiyama, I., Witherington, D. C., Ueno, M., Poutrain-Lejeune, L., & Barbu-Roth, M. (2013). The epigenesis of wariness of heights. Psychological Science, 24 (7), 1361-1367.
Gibson, E. J., & Walk, R. D. (1960). The “visual cliff.” Scientific American, 202 (4), 64-71.
Held, R., & Bossom, J. (1961). Neonatal deprivation and adult rearrangement: Complementary techniques for analyzing plastic sensory-motor coordinations. Journal of Comparative and Physiological Psychology, 54 (1), 33-37.
Karasik, L. B., Tamis-LeMonda, C. S., & Adolph, K. E. (2016). Decisions at the brink: Locomotor experience affects infants’ use of social information on an adjustable drop-off. Frontiers in Psychology, 7, Article 797.
Kohler, I. (1962). Experiments with goggles. Scientific American, 206 (5), 62-72.
Rodkey, E. N. (2015). The visual cliff’s forgotten menagerie: Rats, goats, babies, and myth-making in the history of psychology. Journal of the History of the Behavioral Sciences, 51 (2), 113-140.
Sorce, J. F., Emde, R. N., Campos, J. J., & Klinnert, M. D. (1985). Maternal emotional signaling: Its effect on the visual cliff behavior of 1-year-olds. Developmental Psychology, 21 (1), 195-200.
Stratton, G. M. (1897). Vision without inversion of the retinal image. Psychological Review, 4 (4), 341-360.
