«Department of Psychology, Durham University, South Road, Durham, DH1 3LE, UK To appear in: The Cambridge Encyclopedia of Child Development, 2nd ...»
A class of visual stimuli with huge social importance is the human face. Very young infants, including newborns, already preferentially attend to and orient toward faces (Johnson, Dziurawiec, Ellis, & Morton, 1991), and infants only a few days old look longer at their mother’s face than a stranger’s face. Newborn face processing relies on the very basic visual information available given the low visual acuity and contrast sensitivity at this age, and seems to rely on different mechanisms to those used by older children and adults (Johnson, 2005). These early abilities have been proposed to depend on a specialized sub-cortical pathway including the superior colliculus, pulvinar and amygdala, which may be functional at birth and respond to faces using relatively simple image information present at low spatial frequencies (i.e. features evident even with low visual acuity)..Infants’ face processing is initially quite general, before becoming ‘fine tuned’ by learning. Thus, at 6 months, infants are as good at discriminating two monkey faces from each other as they are at discriminating two human faces (Pascalis, de Haan, & Nelson, 2002). Likewise, EEG recordings at this age do not yet show specific responses to upright (as opposed to inverted) or human (as opposed to monkey) faces, as do those in adults (de Haan, Pascalis, & Johnson, 2002).
Everyday objects can temporarily disappear behind other objects (i.e., become occluded), but humans routinely keep track of them and expect them to re-appear. Although their abilities to follow moving objects using their eyes develops in the early months of life, infants seem at first to be poorly equipped to deal with occlusion. An example is the seminal finding of Jean Piaget (1896-1980) that infants aged below 8 months do not search for an object after it has been hidden from view. Measures of where infants look when an object is hidden (rather than whether and where they search for it) indicate earlier development of abilities to keep track of occluded objects, but this still depends on postnatal development. For example, in studies by von Hofsten and colleagues, infants were able to predict where a moving object would reappear after temporary occlusion at 4 months (see von Hofsten, 2004). EEG measures have related infants’ ability to maintain occluded objects in memory with activity in the temporal lobe (Kaufman, Csibra, & Johnson, 2003). When part of a moving object is occluded, one could interpret it either as a single object, or as two objects that happen to move together.
There is evidence that the assumption of ‘unity’, that the parts all belong to one object, develops at 2-4 months (Johnson, 2004). The origins of infants’ knowledge of objects and physics is a topic much debated in the cognitive development literature.
Depth Another crucial task for the visual system is to compute objects’ positions in 3-D space. The task is not easy because retinal projections are two-dimensional, and do not directly provide depth information. A small object nearby can have a larger retinal projection than a large object far away. Differences between the two eyes provide one basis for judging depth,
disparity information at 12 weeks-of-age. Humans use many additional cues to depth, including patterns of motion produced by rotating objects, perspective, shading, and occlusion (see above). Work by Albert Yonas and his colleagues has documented the development of depth perception based on such cues in the first year-of-life. The pattern of motion generated by a rotating object can be used as a cue to the object’s 3-D shape (the ‘kinetic depth effect’) as early as 8 weeks (Arterberry & Yonas, 2000). Use of ‘pictorial’ depth cues such as perspective, shading and occlusion is evident later, at between 5-7 months (e.g., Yonas, Cleaves, & Pettersen, 1978). Increasingly sophisticated depth judgments based on multiple visual cues emerge in the first year-of-life. This process is likely to depend on learning during infants’ visual and motor interaction with the world.
Visually guided reaching and grasping Human visual and motor systems are tightly coupled, and vision provides a basis both for selecting motor actions (e.g., deciding which object to pick up) and controlling them (e.g., adjusting the hand’s hand orientation so that it can grasp the object). Infants first reach and grasp for objects at around 4-5 months, an ability dependent on the development of motor control of the limbs and of posture, but also on ‘visual’ factors reviewed above including developing perception of objects and of depth. Infants use vision to guide where they will reach as soon as they start reaching, but it is only during their second year of life that they also use vision of their own arm to better control and correct their reach trajectories. By the second or third year-of-life, reaches start to resemble those of adults in being relatively direct to their target and showing a smooth velocity profile. These are indicators of increasingly accurate use of vision for both initial planning and subsequent on-line correction of
including learning to correctly calibrate visual distance to arm length.
Visual development in childhood The most dramatic changes in visual abilities take place in infancy, and infancy is also the period that is most critical for healthy visual development (see below). However, many visual abilities continue to develop through childhood and into adolescence. In general, more basic or ‘low-level’ aspects of image analysis, such as visual acuity and contrast sensitivity, mature early. More complex kinds of analysis, including object and face recognition and use of vision for action, continue to develop long into childhood. The brain’s abilities to make complex visual judgments continue to be fine-tuned by learning and experience throughout childhood.
Global form and motion Studies of sensitivity to global form and motion have found extended development in childhood of abilities to follow contours, and to discriminate coherent movement or patterns embedded in random noise. For example, discrimination is not adult-like until 14 years for either simple up/down dot motion or more complex ‘biological motion’ (moving dots based on the movement patterns made by a human (e.g,. while running; Hadad, Maurer, & Lewis, 2011). These ‘mid-level’ visual abilities depend on integrating form and motion information encoded at lower levels of cortical processing (e.g, combining the motions of multiple dots to extract an overall motion direction). Recent studies have related these developmental changes to improvements in the efficiency with which information is combined (Manning,
reflect changes in cortical connectivity during childhood, as well as developing perceptual expertise.
Objects and faces At even higher levels of the cortical hierarchy associated with object and face recognition, there is similarly evidence for very long visual development through childhood. Young children may be very competent at recognizing common objects in everyday conditions.
However, their abilities to recognize objects in unusual lighting or from unusual viewpoints, or to recognize newly learnt objects from new viewpoints, are still developing late into childhood (Nishimura, Scherf, & Behrmann, 2009). Abilities to tell faces apart based on subtle configural differences are likewise not mature until adolescence (Mondloch, Le Grand, & Maurer, 2002). In adults, these kinds of highly specialized visual processing rely on distinct cortical areas in the temporal lobe, along the ventral visual pathway, which neuroimaging (fMRI) studies have shown also emerge slowly over the course of development, particularly for face processing (Grill-Spector, Golarai, & Gabrieli, 2008).
Development and vision as inference An influential theoretical approach describes perception as probabilistic inference, in which perceptual evidence, which is often uncertain, is interpreted in light of internal models and prior knowledge. This approach goes back to Hermann von Helmholtz (1821-1894), one of the founders of visual psychophysics, and has now been formalized in mathematical models and related to cortical information processing (Clark, 2013)..This account would suggest that children gradually acquire expertise at complex perceptual judgments by improving their
children’s and adults’ abilities to make inferences about 3-D shape based on two sources of evidence (texture and stereo disparity) were compared with model predictions, children made sub-optimal inferences until the age of 12 years. (Nardini, Bedford, & Mareschal, 2010).
Likewise, children’s propensity to interpret ambiguous 3-D shapes based on the prior assumption that the light is most likely to be coming from above is still developing at 10 years (Thomas, Nardini, & Mareschal, 2010). How the developing brain learns to deal with uncertain sensory information and so make ‘optimal’ perceptual inferences is a topic of current research. The perceptual inference contrasts with other influential theoretical frameworks, such as the ‘ecological’ approach pioneered by James Gibson (1904-1979) and Eleanor Gibson (1910-2002), which emphasises the rich sensory information available during naturalistic tasks.
Vision and action Another domain showing marked development throughout childhood is visually guided action, including manual tasks (pointing, reaching), balance, locomotion and navigation. The gradually developing expertise in visual recognition (e.g., of faces), supported by the ventral stream of visual processing, is paralleled by developing expertise in planning and executing visually guided actions, supported by the dorsal stream.
Atypical visual development As has been described, healthy visual development involves the acquisition of many different perceptual abilities. There are many points at which processing can be disrupted, from the eye (e.g., lens, retina), to the brain. Neonatal visual processing is very immature, and both the
movements, have to develop postnatally. This development usually proceeds normally, but is vulnerable to disruption, especially in absence of normal visual experience.
The eye Clearly, disorders of the eye can impair visual function. Most common are refractive (focusing) errors, which can be corrected with glasses. Rarer conditions include congenital cataracts and genetic conditions such as nightblindness affecting cells in the retina. Babies born preterm are at risk of retinopathy of prematurity, in which abnormal development of blood vessels damages the retina. Other developmental eye conditions include congenital or infantile glaucoma (elevated intraocular pressure, which can lead to impaired vision and damage to the eye). When it is possible to correct vision at the level of the eye, it is crucial to do so early to ensure normal development of eye control and the visual brain, as described below.
Eye and brain: Strabismus, nystagmus, and amblyopia The control and feedback mechanisms between brain and eye can fail to develop correctly for guiding eye movements. In strabismus (‘crossed eyes’), the directions of the two eyes are misaligned. Other primary vision problems (e.g., congenital cataract, or much more commonly, far-sighted refractive errors) are risk factors, reflecting the developing system’s need for a clear visual signal to calibrate itself correctly. In early-onset or congenital nystagmus, the ability to hold the gaze steady does not develop normally, and there is uncontrolled back-and-forth movement of the eyes. Reduced vision is also a risk factor for nystagmus, although it can also have a neurological cause in either genetic conditions or acquired injury affecting the developing eye-movement system.
In amblyopia, the brain’s processing of the visual information provided by the eye does not develop normally. The great plasticity of the developing visual cortex means that if, early in life, cortical neurons receive much better information from one eye than the other, they develop to make use of the signal from the ‘good’ eye while discounting information from the other eye. Therefore, early visual problems at the level of the eye, such as cataracts, poor focus, or strabismus, can lead to amblyopia: abnormal development of visual cortex, leading to lasting vision impairment. Because there is much less cortical plasticity (potential for reorganization) in later life, amblyopia typically remains even if the problem is subsequently dealt with at the level of the eye. For this reason, where possible early interventions are used to improve vision from a weaker eye (e.g., focal correction, cataract surgery), and to encourage the brain to use the signal from both eyes (e.g., patching treatment to promote use of the weaker eye). However, new research suggests that there may be more scope for lifelong cortical plasticity allowing treatment of amblyopia than was previously recognised (Bavelier, Levi, Li, Dan, & Hensch, 2010).
Deprivation and brain plasticity Well-known, clinically recognized visual problems associated with early visual deprivation (e.g., congenital cataract) include lasting impairments in visual acuity and contrast sensitivity.
These are associated with an under-representation of the signal from the affected eye at the initial levels of cortical visual processing. However, visual deprivation is also associated with impairments in higher-level functions such as coherent motion processing, shape and face recognition. These functions, too, need normal early visual input to develop, and patients who had bilateral cataracts removed only in late childhood or adulthood show marked deficits in them. Interestingly, infants who had cataracts removed at an age younger than those at which