«Department of Psychology, Durham University, South Road, Durham, DH1 3LE, UK To appear in: The Cambridge Encyclopedia of Child Development, 2nd ...»
Department of Psychology,
Durham University, South Road,
Durham, DH1 3LE, UK
To appear in: The Cambridge Encyclopedia of Child Development, 2nd Edition. Hopkins,
B., Geangu, E. & Linkenauger, S. Cambridge University Press.
M Nardini, Vision p. 1
In humans with healthy vision, sight provides a major source of information about the world.
Newborns have some rudimentary visual abilities (e.g.,,they orient toward large, highcontrast stimuli), but most visual abilities develop significantly postnatally. Some aspects of vision continue to develop well into childhood and even adolescence. This development usually unfolds in a predictable sequence, but crucially depends on having normal visual experience.
A brief sketch of the anatomy of the human visual system will provide a context for understanding its development. Light focused by the cornea and lens falls on the retina, at the back of the eye, where light-sensitive cells (photoreceptors) convert patterns of light into neural signals. Three classes of cone photoreceptors sensitive to different wavelengths of light provide a basis for color vision and seeing fine detail, while rod photoreceptors are specialised for seeing in low light. After initial processing by retinal neurons, signals from the retina are relayed to the brain by the optic nerve. A major pathway is one to the primary visual cortex (V1) in the occipital lobe via the lateral geniculate nucleus in the thalamus (LGN). From V1 onward, visual information is processed by a complex hierarchy of cortical visual areas. Two major pathways are the ‘ventral stream’, subserving recognition of objects or faces, and the ‘dorsal stream’, providing information about spatial layouts to guide actions such as reaching and grasping. The function of this network largely develops postnatally.
Other, earlier-functioning visual pathways depend on sub-cortical rather than cortical processing. These networks include structures such as the superior colliculi and oculomotor nuclei. These pathways are involved in simpler, more reflexive visual functions, and underpin many early-developing visual responses such as the vestibulo-ocular reflex (which helps to M Nardini, Vision p. 2 keep the eye’s image steady when the head moves). Other important eye movements such as saccades are controlled by signals from both cortical and sub-cortical brain networks, as described below. Thus, developing a normal visual system involves setting up complex interactions between eyes and brain, and between brain structures.
Key components of the visual system develop prenatally, including the pathway from the retina to the LGN. In utero, before any visual experience, retinal ganglion cells fire spontaneously, generating ‘waves’ of activity across the retina. This spontaneous activity is thought to play a crucial role in organizing the synaptic connections between retina and LGN (Katz & Shatz, 1996). Following birth and the onset of vision, the neural circuits comprising the visual system continue to be sculpted by visual experience. This entry will first describe normal visual development in infancy and childhood, followed by disorders of visual development.
Visual development in infancy Human vision includes many different abilities from recognition of faces to guidance of accurate movements. A basic aspect of visual sensitivity is visual acuity: the ability to resolve fine detail, as tested (in adults) by an optician’s letter chart. Limitations in the ability to see details would provide a bottleneck for any further visual analysis that depends on such details. A closely related function is contrast sensitivity: the ability to distinguish a pattern from its background based on differences in lightness. Good contrast sensitivity entails being able to see patterns based on subtle lightness differences. Behavioral and brain measures (EEG) show that newborns’ visual acuity and contrast sensitivity are many times lower than those of adults, but that both normally improve rapidly during the early months of life, reaching adult-like levels by 3-4 years. A standard behavioral measure of infant visual acuity
measures infants’ preferences to look at increasingly fine patterns printed on cards as compared with uniform gray. Newer methods can measure acuity automatically using eyetracking and on-screen stimuli (Jones, Kalwarowsky, Atkinson, Braddick, & Nardini, 2014).
EEG measures of acuity include the ‘sweep VEP’ (visual evoked potential), in which brain responses to progressively finer spatial patterns are recorded.
The rapid, early development of acuity and contrast sensitivity is not well explained by optical changes in the eye (e.g., ability to maintain focus), but depends on changes in the retina, developing brain connectivity, and improvements in neural information transmission.
At birth, cone photoreceptors are immature in their morphology, inefficient at capturing light, and not densely packed in the fovea (the central region of the retina that provides the highest visual acuity). Morphological maturity and spatial re-arrangement of cones in the first two years of life provide one basis for improving visual function. Other crucial factors are experience-dependent sculpting of the neural circuits between retina, LGN and V1, and improvements in the efficiency of these circuits via myelination.
As these brain changes depend crucially on having normal visual experience, when vision is limited during infancy, for example because of a congenital cataract, visual function is compromised, even once the original impairment (e.g., cataract) is removed. This condition, amblyopia, has its basis in the brain, as shown using animal models in the Nobel prize-winning work by David Hubel (1926-2013) and Torsten Wiesel (Wiesel, 1982). In absence of detailed visual input from one eye, neurons in visual cortex come to be strongly biased to process information only from the other eye. Animal models have shown in detail how the plasticity of these initial visual circuits is greatest during ‘critical’ or ‘sensitive’
in early and later life, see sub-section on Atypical visual development below.
The eye and eye movements Although humans have a wide visual field, only a very small region at its center provides very fine vision. For this reason, the eyes are constantly on the move, sampling the environment. These rapid, jerky eye movements are called ‘saccades’. Other kinds of eye movements provide stabilization to account for movement (e.g., continuous tracking of a moving target is termed ‘smooth pursuit’). Newborn control of eye movements is quite basic, and setting up precise eye movement control is a key challenge for the visual system. In the first months of life, normal visual development includes development of the abilities to converge both eyes to hold an object steadily in view, to track a moving object smoothly, and to shift the gaze flexibly from one target to another. These abilities depend on different but interrelated brain networks, including dedicated sub-cortical pathways and nuclei distinct from the main cortical visual pathway via LGN and V1. For example, the smooth pursuit network includes areas V1 and MT in the occipital lobe, the frontal eye fields in the frontal lobe, and sub-cortical structures including the pontine nuclei and the cerebellum. A hallmark of normal visual development is an increasing ability to drive eye movements not only by relatively inflexible sub-cortical mechanisms, but also by flexible cortical control based on more detailed image analysis. For example, infants aged below 3-4 months of age tend to get ‘stuck’ on a target and are unable to disengage from it even when another attractive target appears alongside (Atkinson, Hood, Wattam-Bell, & Braddick, 1992). The ability to disengage is an indicator of developing cortical control, and can be delayed in infants born prematurely or with perinatal brain injury. The assessment of this ability can therefore provide a useful indication of brain development in ‘at-risk’ groups.
Visual acuity and contrast sensitivity describe basic limits on spatial information available for the visual system to use. Related measures are chromatic sensitivity in the domain of color discrimination, and critical fusion frequency in the time domain (the latter describes how rapidly a pattern can flicker between two levels of lightness or color for the flicker still to be perceived, rather than ‘fused’ into one). Importantly, learning to see is not just a matter of overcoming these basic limitations, which one might liken intuitively to the progressive sharpening of a blurry image. There is still all the difficult work of visual analysis to be done, a process carried out by a complex network of cortical visual areas starting with V1 in the occipital lobe and taking up a large portion of the human brain. Neurons in visual areas at lower levels of this hierarchy are sensitive to simple, small, local image properties, such as edges or corners; those higher up to more extended properties such as contours stretching over the visual scene. At the highest levels, neurons are sensitive to complex stimuli including faces and objects. Are these kinds of cortical processing present at birth or do they develop postnatally? If so, how?
Form, motion, and disparity These questions were addressed in a series of landmark studies by Atkinson, Braddick and their collaborators beginning in the 1970s (Atkinson, 2000). They used both behavioral (preferential looking) and EEG measures to look for signatures of cortical visual processing related to key properties of V1 known from animal models: orientation, motion direction, and binocular disparity. Findings from these studies are summarized in Figure 1. Their results showed that all three kinds of cortical visual processing are almost entirely absent at birth,
relatively fast rates of flicker, signalling that something is changing or moving in the visual scene, they cannot resolve the direction of motion until age 2-3 months (Wattam-Bell, 1991).
This shows that cortical mechanisms for interpreting the direction of motion normally develop in the early months of life. The same is true for orientation (direction of a static pattern), and binocular disparity (differences in the two eyes’ images, which provide one basis for seeing in depth, see below). All these normally develop postnatally, at different ages (see Fig. 1). A different, and more basic kind of motion sensitivity based on sub-cortical pathways underlies the optokinetic nystagmus (OKN) reflex, in which the eyes follow large patterns moving horizontally from birth.
Figure 1. Summary of behavioral and EEG results from studies by Atkinson, Braddick and colleagues.
Redrawn from Atkinson (2000).
These basic sensitivities to orientation, motion and disparity are linked to properties of V1 neurons, which carry out initial image analysis. Each V1 neuron only responds to simple image properties in a very small part of the visual field. Neurons higher up in the cortical hierarchy respond to more complex properties over larger parts of the visual field. Two aspects of this ‘mid-level’ visual processing are sensitivity to ‘global form’ and ‘global motion’ (as distinct from ‘local form’ or ‘local motion’). Determining the lines of extended edges is an example of global form processing, while segregating coherent parts of objects moving against a background is an example of global motion processing. Abilities to process global form or motion can be tested by comparing behavioral or brain responses to ‘coherent’ as compared with ‘incoherent’ patterns (see Fig. 2). In these patterns, the component local elements are the same so the patterns should be similar from the point-of-view of V1 neurons, but the global organization is different.
Global form and motion processing are expected to build on local processing. In line with this expectation, they are first seen in infancy, via behavioral and EEG recordings, a little later than local processing at around 4-6 months. Interestingly, while EEG responses to globally organized form and motion are evident at 5-months-of-age, the topography of these responses as recorded over the scalp are very different to those recorded in adults (WattamBell et al., 2010). This difference indicates that there is major reorganization of cortical visual processing between the first emergence of these kinds of sensitivity and their final adult state. The nature of this reorganization in both function and connectivity of cortical visual areas is not yet well understood and remains a topic of current research.
distributed in the same way in both stimuli, but only in the coherent case do they follow a larger (‘global’) organisation.
Color Normal color vision depends on three kinds of retinal photoreceptor, which respond to different wavelengths of light: ‘L’, ‘M’ and ‘S’ (long-, medium- and short-wavelength) cones, corresponding to colors seen as red, green and blue. Newborn color vision is limited in showing low sensitivity to chromatic contrast, but by two months there is evidence for discrimination for red-green contrasts, mediated by L- and M-cones. Discrimination for blueyellow contrasts, mediated by S-cones, seems to develop later, after 4 months. This early development is likely to be driven by the maturation of cone photoreceptors, which change in shape to become more efficient at catching light, together with the development of cortical information processing (sensitivity to visual form or motion also depending on this change).
Infants’ earliest responses to color, however, may be based on a sub-cortical visual pathway via the superior colliculi which does not receive S-cone input.