Eleanor Gibson & Richard Walk, 1960

 

 

AIMS:  Depth perception is our ability to perceive how close or far an object is from us - perception being the process of understanding visual information.  It is just one element of our visual capabilities; but it is an essential perceptual ability to have as we negotiate our way around our world. If we had no depth perception, we would probably not survive very long – eg: it would be extremely hazardous trying to ascend of descend stairs or jump off a wall without depth perception.

Nativists (people who support the Nature side of the Nature vs Nurture debate) believe that we are born with certain capacities such as the ability to perceive depth. They believe that these abilities may not all be functioning properly when we are born but that the process of maturation determines the development of these capacities. (Eg: the optic nerve does not develop a myelin sheath around it - for faster neural transmission - until the child is around 4 months of age.) Learning is of little or no importance in the development of these abilities. Empiricists or Nurturists (people who support the Nurture side of the Nature vs Nurture debate) on the other hand believe that we acquire our skills through our experiences; our abilities being the product of our ability and willingness to learn. Interactionists believe that our abilities are the product of innate abilities interacting with environmental factors.

When we are born the nervous system has all appropriate components, however it is immature - it is about half the size of an adult. The optic nerve, for example is obviously shorter than it will be when it is adult sized; but it is also narrower as it does not have the necessary myelin sheath that ensures good transmission of information.

Nativists would assume that depth perception would be an innate characteristic; Empiricists would assume that depth perception is acquired in the time before we become independently mobile; Interactionists would assume that depth perception is the product of the developing visual system - eg: the myelin sheath around the optic nerve is thought to be fully developed by the age of 4 months) - and experience – i:. from the time we are born we are exposed to various complex and intriguing stimuli, such as faces, which have subtle cues about depth – such as shadowing).  

Although most infants start to demonstrate some independent locomotion by the age of 6 months, many species are able to demonstrate movement from the time that they are born.  Gibson & Walk decided to use not only human infants in their research, as that would be inconclusive as to whether Nativist, Empiricist or Interactionist arguments were correct. By using animals, such as kids (infant goats), lambs, chicks and cats they were able to investigate if cliff avoidance behaviours were evident from birth in these species. Other animals such as aquatic turtles were also used, to investigate if those species whose environment does not necessitate a particularly strong need for depth perception were less likely to be able to demonstrate it.

Early studies had involved rearing animals in the dark and seeing whether their lack of visual experience affected their ability to perceive distance normally. Karl Lashley & J T Russell  (1934) had reared rats in the dark and found they could still jump the correct distance on to a platform. Gibson & Walk criticised this on the grounds that, because the rats had to be trained to jump, they could simply have learned to jump the distance correctly during this training.

Gibson & Walk aimed to investigate if infants could discriminate depth by the time they were able to move independently.  They wanted to assess whether an infant’s perception and cliff avoidance behaviours were an innate characteristic.

 

PROCEDURE (METHOD):

At Cornell University, the researchers assembled the ‘visual cliff’ apparatus in a laboratory. The visual cliff consisted of a large glass sheet which is supported a foot or more above the floor with patterned material directly beneath the glass on one side, and several feet below it on the other. Chequered patterns on the material underneath the glass gave visual cues that one side was ‘shallow’; the other side was ‘deep’. The glass top not only kept the participant safe but also meant that any other non-visual clues were not available via other senses.

36 infants from the age of 6-14 months - who was available to them - were placed on the centre board of the visual cliff. All the infants were able to crawl (independent locomotion). Each child was then observed to see if they would crawl to their beckoning mother (cross onto the deep side), stay on the centre board or crawl onto the shallow side (away from the beckoning mother). The child’s mother also called to them from the shallow side.

If they would not crawl across the ‘cliff’, this would suggest that their depth perception was intact by the time they could crawl.

Non-human animals – eg: kittens, rats, chicks, goats, puppies, pigs and aquatic turtles - were also tested on the visual cliff. They were also tested at the age at which they demonstrate locomotion - ie:-.

• Chicks - 1 day old
• Lambs - 1 day old
• Kids (infant goats) - 1 day old
• Rats - 4 weeks old
• Kittens - 4 weeks old and ‘reared normally’
• Kittens - 4 weeks old but ‘reared in the dark for 27 days’

The chicks, kids and lambs were particularly interesting because they are precocial - they can stand and show some mobility and, thus, be tested within 24 hours of birth. Thus, it was theorised, they would have had no opportunity to learn to perceive depth. Having depth perception at birth would be considered adaptive - as it would be essential for the inexperienced precocial neonate to survive.

A further test was carried out on the kids and lambs. An adjustable cliff was set up so that the apparently shallow surface could be lowered once the animal was on it. This was to test the animal’s response to visual cues that the surface was suddenly falling away.

In a further condition, the chequered pattern beneath the glass was replaced by a uniform grey surface. This was to test whether it was the pattern that allowed the participants to perceive depth.

A number of measures were put in place to control extraneous variables - eg: they lit the chequered surfaces from below the glass to prevent reflections in the glass.

76% of the aquatic turtles crawled off onto the shallow side but a number did venture on to the deep side.

In the condition with the lowered surface, all of the animals adopted an immobile defensive posture when the surface dropped more than 12 inches below the glass. The animals did not adapt when the procedure was repeated a number of times but continued to freeze.

When the chequered pattern was replaced with uniform grey, no animals showed any preference for one side over the other.

 

CONCLUSIONS

All of the species tested, including humans, showed intact depth perception by the time they could move independently. In some species this was within a few hours after birth. This suggests their ability to perceive depth was present at birth, supporting the Nativist viewpoint.

The fact that no species showed a preference for the shallow side over the deep side in the ‘uniform grey’ condition suggests that the innate mechanisms for depth perception involved interpreting changes in patterns indicating depth.

The large minority of aquatic turtles that chose the deep side suggests the turtle has poorer depth discrimination than other animals. Gibson & Walk suggest that in its natural habitat does not really pose it with the ‘occasion to fall’.

“The survival of a species requires that its members develop discrimination of depth by the time they take up independent locomotion, whether it be at 1 day (the chick and goat), 4 weeks (the rat and cat), or 6-14 months (the human infant).  That such a vital capacity does not depend on possibly fatal accidents of learning in the lives of individuals is consistent with Evolutionary Theory.” - ‘The Visual Cliff’ in Scientific American #202

 

EVALUATION (CRITICISMS):

The visual cliff was an outstanding design, enabling depth perception to be tested safely, with all sensory cues other than visual to be eliminated. Gibson & Walk’s use of it was arguably the first successful procedure to be used for measuring depth perception in infants.

The infants’ movements were an easily identifiable measure of their perception.

Previous research into depth perception in animals did not easily facilitate the findings being applied to humans. However, by testing a range of species besides humans, the consistency of their findings across the range shores up the credibility of Gibson & Walk’s conclusions.

There are, however, a number of flaws in the methodology.

Firstly, the sample size of 36 infants was rather small and the age range  (6-14 months) rather large. Since it can be argued that human babies have had ample time to learn depth perception by the time they can crawl, the fact the vast majority did not venture on to the deep side cannot be used to support the Nativist position. However, Gibson & Walk insist they drew their conclusions by considering the data from all the species they tested.

While the direction of the movement of the participants - both human and animal - can be said to be an indication of depth perception, none could tell the researchers they were motivated to avoid a drop. Gibson & Walk assumed this from their observations.

Then there is the ethical issue of the babies becoming distressed by their mother appearing to beckon them to fall off a cliff. Also simply being perched on a centre board next to a drop may have been frightening.

And, of course, the babies neither gave consent nor had the right to withdraw - though their mothers did!

In an attempt to get around the issue of testing infants before they can crawl, Sandra Scarr & Philip Salapatek (1970) simply wheeled young infants across the deep side in a trolley; the infants showed no wariness. Joseph Campos, Alan Langer & Alice Krowitz (1970), in a slightly more sophisticated study, placed 2, 3.5 and 5-month-old infants on the shallow and deep sides of the visual cliff and measured their heart rates. Even the youngest had a drop in heart rate, showing interest when placed on the deep side. They were also less likely to cry and more likely to pay attention to what was underneath them; they certainly seemed unafraid. There were no such changes when they were placed on the shallow side. Campos, S Hiatt, D Ramsay & M Svejda repeated the study in 1978 and found again that infants between 2 and 5 months placed on the deep side actually had slower heart rates, suggesting they were interested and certainly were not frightened. Andrew Schwartz, Joseph Campos & Edward Baisel (1973) found that 5-month-olds placed on the deep side showed no increase in heart rate but 9-month-olds did. While these findings might appear to support at least the Interactionist view, it could be that the 2-to-5-month-olds could perceive depth to some degree but did not yet understand the implications of what they saw on the deep side for their safety.  Interestingly, in contrast to Gibson & Walk’s infants patting the glass on the deep side but still not daring to venture on to it, Karen Adolph & Sarah Berger (2006), testing younger babies on the visual cliff, found that, when they slipped off the centre board on to the deep side, they realised that they didn’t fall and that the glass was safe.

James Sorce, Robert Emde, Joseph Campos & Mary Klinnert (1985) had the mother put on either a facial expression of happiness or one of fear on the other side of the cliff. When the mother put on the ‘happy face’, the babies checked the cliff but crossed; when she put on a fearful expression, they didn’t.

Intrigued by footage of earlier studies in which even the youngest of babies braced themselves before touching the shallow side, David Witherington, Joseph Campos, David Anderson, Laure Lejeune & Eileen Seah (2005) challenged whether the visual cliff was actually measuring depth perception. In their study, the first group of 20 infants were experienced at crawling but were not yet walking while a second group of 20 infants had just begun to walk. The researchers found that the older infants in the second group were more wary of the deep side than the younger in the first group. They concluded that what is really going on in visual cliff studies is that the infant is learning to associate the physical experience with the visual environment and that new learning has to take place when the world is viewed from a new perspective - ie; walking.

In non-visual cliff research on depth perception, T G R Bower, J M Broughton & M K Moore (1970) found credible evidence that infants just 6 days old have some depth perception. The researchers showed the infants 2 objects: a large disc that approached to within 20 cms of them and a smaller disc that came to within 8 cms of them. The expectation was that, if the babies lacked any depth perception, their response to the larger disc stopping further away would be the same as the smaller and closer one because they both would form the same retinal image. In fact, the researchers had to abandon the experiment without testing all the infants because the babies were so upset by the smaller disc getting so close to them, rotating their heads and pulling away. Albert Yonas & Cynthia Owsley (1987) showed 2-3-month-old babies a video of an object which appeared to be moving directly towards them. The babies showed avoidant behaviour - blinking, flinching, moving their head to one side, etc. Earlier Yonas (1981) compared 2 groups of 6-week-old babies - the first having been born on time, the second 4 weeks late. The second group were significantly more likely to respond to a looming object with an avoidance reaction. Yonas argued that this supported the Nativist view as both groups had had the same environmental exposure.

Most depth perception is based on monocular cues such as:-

• Relative size - smaller objects appear further away
• Texture gradient - there is more detail in objects that are closer
• Occlusion (overlap) - if one object partially blocks another, it appears to be closer
• Linear perspective - parallel lines appear to converge in the distance
• Motion parallax - as we move, objects closer to us appear to move more quickly than do those further away

Claes von Hofsten, Philip Kellman & Jorma Putaansuu (1992) demonstrated the use of motion parallax in 3-month-old infants with the habituation method - ie: they showed the infant a display until the infant habituated - got used - to it. The infants were moved about in a chair while being habituated to a display of 3 rods. These were at the same distance but the middle one was moved in synchrony with the infant to create a motion parallax effect. The infant was then shown 2 further displays, one of 3 rods at the same distance from the infant and the other with the middle rod further away, matching the motion parallax effect. The infants showed more interest in the 3 equidistant rods - a new display to them, demonstrating that they had the ability to use motion parallax.

However, Yonas, Carl Granrud, Martha Arterberry & Brenda Hanson (1986) provided evidence that the ability to respond to depth cues in pictures emerges rather late. The ability to respond to occlusion emerged at around 6 months while texture gradient and linear perspective emerged at around 7 months. Previously Yonas & Granrud (1985) had found that 5 month-old infants appeared unable to use the pictorial cues from occlusion/overlap to distinguish one object being nearer to them than another while 7-month-olds were able to. In 2001 Yonas, Melissa Farr & Ann O’Connor demonstrated that the depth cues provided by shadows could be understood by older babies (average age 30 weeks) but not by younger babies (average age 21 weeks). The infants, one eye covered to removed binocular information for depth. were shown 2 equidistant toys, one with added shadows - th e older babies reached for the shadowed toy (appearing closer) but the younger ones did not. Gavin Bremner (1994) proposed that the ability to interpret the kind of dynamic cues Bower et al presented appeared earlier than the ability to use static pictorial depth cues.

In work with adults M J Sinai, T L Ooi & Z J He (1988) tested their ability to judge distance up to 7 metres. They found that, when the ground was even, participants could use texture as a cue and judge distance very accurately. However, when other stimuli, such as a ditch, were put in their way, the accuracy of their distance estimates declined.

In judging short distances, such as those involved in the visual cliff study, binocular cues can be used as well as monocular clues like pattern. Binocular cues include stereopsis - analysis of the differing information entering each eye. Benjamin Backus, David Fleet, Andrew Parker & David Heeger (2001) demonstrated this when they found areas of the brain known to be sensitive to differences between 2 sources of information were seen to be active during depth perception tasks. This is important because of binocular cues such as retinal disparity (the closer an object is, the more disparate the images in the 2 eyes will be).

Grazyna Tondel & Rowan Candy (2007) presented 2-5-month old babies with the the image of a fast-moving clown. Most of the babies could track the clown, even when it moved at a speed of 50 cms per second. Tondel & Candy claimed this supported the Nativist view.

However, Francesca Pei, Mark Pettet & Anthony Norcia (2007), using a range of different textures and patterns, found that, although human infants can use crude patterns like the squares in the visual cliff, they cannot detect more subtle differences in texture in the way adults can.

Video extract from the Visual Cliff study

RESULTS (FINDINGS):

All of the 27 infants who moved off the centre board crawled out on to the shallow side at least once. Only 3 attempted to crawl on to the ‘deep’ side (cliff side). (Some of the infants did back on to the deep side accidentally when negotiating the centre board.) Many of the infants crawled away from the mother when she called to them from the ‘deep’ side; others cried when she stood there because they could not get to her without crossing the deep side.

The infants often patted the glass on the deep side with their hands so they knew there was a solid surface; but the appearance of a drop was enough to stop them crawling out onto it.

The chicks, at an age of less than 24 hours, would always hop off the centre board on to the shallow side, rather than the ‘deep’ side. The kids and lambs never stepped on to the ‘deep’ side, even at 1 day old. Chicks, kids and lambs tended to move to the ‘safe’ side of the centre board. When placed on the deep side, they ‘froze’ - front legs rigid and hind legs limp.

Rats, which depend upon their whiskers to navigate rather than visual cues, showed little preference for the shallow side so long as they could feel the glass with their whiskers. When the centre board was placed higher than their whiskers, they nearly always descended onto the shallow side (95-100% of the time).

Kittens - although they rely on their whiskers, also use sight as they are predatory - at 4 weeks old showed preference for the shallow side and froze when placed onto the deep side or circled back to the centre board.

Kittens which had been reared in darkness for their first 27 days of life crawled onto the shallow and deep side equally. When placed on the deep side, they demonstrated similar behaviours to if they had placed on the shallow side. They did not freeze or circle back like the ‘normal’ kittens. After this initial research, these kittens were kept in ‘normal’ lighting conditions. They were tested daily on the visual cliff and by the end of 1 week the ‘dark reared’ kittens demonstrated similar behaviours to kittens who had been reared in the light - ie: almost unanimous preference for shallow side.