The one researcher who has had the longest—and many would say the most profound—impact on the field of infant cognition and perception is Piaget. Originally a biologist, Piaget developed a theory of cognitive development by observing his own children's behavior on certain tasks during infancy and childhood (e.g., Piaget, 1927, 1936/1952, 1937/1954). For many years, psychologists in the United States disregarded Piaget's theory because his research methods were considered imprecise and his ideas about cognition were in conflict with the be-haviorist's zeitgeist of the day. However, that position began to change as comprehensive reviews of Piaget's theory became available in English (e.g., Flavell, 1963; Hunt, 1961). As the reader will see, the modern infant cognition researcher often uses Piaget's theory and observations as a springboard for further ideas and research.
Piaget's (1937/1954) view of infant development is that infants develop an understanding of the world—that is, an understanding about objects, space, time, and causality—by interacting with the environment. Borrowing from the field of biology, Piaget (1936/1952, 1937/1954) believed that infants develop through the processes of assimilation and accommodation. Piaget also believed that development is stagelike and discontinuous. Furthermore, the infant, according to Piaget, is as an active learner who is motivated to learn about the world; but cognitive development, like other aspects of development, represents an interaction between maturation and learning.
Piaget specified four major periods corresponding to different ages of the developing child. The first period of Piaget's (1937/1954) theory of cognitive development, the sensorimotor period, describes infants from birth to around 18-24 months of age, or about the age that language first appears. During this period, infants are thought to go through six stages, starting from interacting with the world strictly with innate reflexes (Stage 1) to using mental representations for acting on the world (Stage 6). Topics examined by Piaget include the development of infants' understandings of time, space, causality, and the permanence of objects.
Many modern developmental researchers agree with Piaget's view that the child is an active learner, but disagree with his view that development is discontinuous, or stagelike. Theoretical positions, such as information processing and the connectionist view of development, are similar in some ways to Piaget's view in that both emphasize the role of learning and experience to help explain developmental changes and both can be considered constructivist theories. Like Piaget, both assume that developmental change is a building-block process that starts small and gradually becomes more elaborate or sophisticated. Other more nativist views believe that Piaget was much too conservative about infants' developing abilities; they claim he underestimated the ability of the infant. Some modern researchers believe that because Piaget's method of research was not truly experimental, his findings were not generalizable. His findings are sometimes criticized for erroneously focusing too much on the child's competence at a very specific task that may or may not reveal the child's true understanding of the world (e.g., Baillargeon, Spelke, & Wasserman, 1985; Bower, 1974). Nevertheless, Piaget is revered today by many infant cognition and perception researchers and is appreciated for his ingenuity and his insights into the mind of the infant (see Flavell, 1963).
Gibson's Ecological Theory
Not all theories of development rely so much on the developing mind of the infant. For example, Gibson's ecological approach to the study of infant perceptual development places emphasis on the environment and infant's abilities to detect important information from the world. Her view is based primarily on two key issues: (a) the infant's ability to discover new affordances—ways upon which an environment lends itself to be acted, and (b) the infant's ability to differentiate— parse out invariant information from the world. As infants act in the world, they differentiate information in their environment and discover affordances. With this new understanding of the world around them, their actions in the world change.
According to Gibson, perception and action are closely related for the infant, and much Gibsonian research examines that relationship (Gibson, 1995). For example, it has been found that an infant may tumble down a slope the first couple of times he or she approaches such a surface. Soon the infant begins to understand that one affordance of the slope, compared to other surfaces, is that it may cause tumbling (Adolph, Eppler, & Gibson, 1993a). Another example is Gibson and Walk's (1960) classic work on infants' perception of depth, known as the "visual cliff" experiment. In this study, it was found that infants would not crawl across a table that appeared to have a drop-off, or cliff, in the middle. Gibson and Walk saw this as evidence that infants of crawling age had enough experience with depth to know that it could afford danger. (For a more detailed discussion of this theory and related research, see the review by Adolph, Eppler, & Gibson, 1993b).
Another theoretical view that emphasizes the close relationship between perception, cognition, and action is dynamical systems. In a recent set of books, Smith and Thelen (1993) and Thelen and Smith (1994) attempt to unify recent advances in dynamical systems theory with research in developmental neuroscience and behavioral development. They argue that development can best be understood in terms of complex nonlinear systems that are self-organizing. Developmental changes tend to be described in the language of physical nonlinear systems—that is, attractor states, phase transitions, and stability versus fluctuations of the system. This view, with its emphasis on mechanisms of change, is clearly opposed to nativist explanations. Although it has been applied most successfully to issues of motor development, its advocates are attempting to apply it to perceptual and cognitive development as well. As we shall see, it also has important elements in common with connectionist modeling.
A persistent theoretical debate throughout the entire history of developmental psychology has been nativism versus empiricism. Nowhere is this debate more apparent than in modern-day infant perception and cognition. A chief spokesperson for the nativist position is Spelke. According to Spelke (1985), infants are endowed by nature with capacities to perceive and understand objects and events in the world. This core knowledge includes an understanding of partially and fully occluded objects; the ability to reason about physical properties of objects, such as continuity and solidity; an understanding of number; and knowledge of physical causality. Spelke and others, most notably Baillargeon, have argued that human infants are more competent than others (such as Piaget) had believed. To bolster their claims, they have provided considerable evidence based upon ingenious variations of methods involving infant habituation and visual preference. In some of these variations, infants must not only perceive events involving one or more objects, but must also make inferences when a portion of those events is hidden behind an occluder.
Needless to say, this viewpoint has been controversial. Many believe the nativist assumptions about the competencies of young infants are unwarranted. In fact, they question whether such assumptions can even be considered explanations. Recent debates on the pros and cons of nativism have appeared in the literature between Spelke (1998) versus Haith (1998) and Baillargeon (1999) versus Smith (1999). Furthermore, empirical research on some of these topics is beginning to show that simpler explanations can account for the apparent cognitive sophistication proposed by the na-tivists. Some of the research both for and against the nativist viewpoint is described in later sections of this chapter.
The connectionist modeling approach stands in sharp contrast to the nativist approach. The most complete description of the application of connectionism to development has been published in a recent book entitled Rethinking Innate-ness (Elman et al., 1996). Whereas nativism assumes infants come prewired with certain core knowledge, connectionists reject this form of innateness. They argue that at all levels— molecular, cellular, and organismic—interactions occur between organisms and the environment. A more appropriate meaning of innateness is that the outcome is constrained to some extent at each of those levels. These constraints operate on the type of representation, the architecture, and the timing of the developmental process that is being considered. Comparisons—sometimes real, sometimes metaphorical— are made between the structure of the brain and computerized connectionist networks. So, for example, just as brains include neuronal synapses and specific areas, connectionist models include patterns of connections and types of units or levels. Connectionist models also include nonlinear learning rules that may lead to emergent, stagelike properties and thus are quite compatible with a dynamical systems approach. Many early models were developed to counter the prevalent nativism of linguistic theory. Modelers tried to demonstrate that what some assumed were innate rules of language could be approximated by connectionist models through experience. Connectionism has spread to simulations of infant perception and cognition. New models are appearing on infant categorization (Mareschal & French, 2000), object permanence (Munakata, McClelland, Johnson, & Siegler, 1997), speech perception (Schafer & Mareschal, 2001), and rule learning (Shultz & Bale, 2001). The adequacy of these models is still being hotly debated (Marcus, 1999a, 1999b) but there is no doubt that these models are presenting a challenge to the view that infants possess innate knowledge structures.
Developmental neuroscience is an area that has grown substantially over the last 10 to 20 years. Like dynamical systems and connectionist modeling, its users attempt to make links between development of the brain and development of behavior associated with perception and cognition. Whereas previous approaches to ties with brain development have been somewhat metaphorical, developmental neuroscientists attempt to measure brain development directly and then relate it to cognitive development. However, finding the answers to questions about brain development and cognitive development in infants is not such a straightforward business. Procedures that may be useful for studying the link between brain and cognition in older children may not be appropriate for studying this link in infants. For example, PET (positronemission tomography) and fMRI (functional magnetic resonance imaging) methods may work fairly well with children and adults, but these are considered too intrusive for use with infants (M. H. Johnson, 1997). The methods most often used to study the developing brain in infants are EEGs (electroen-cephalographs), ERPs (event related potentials), and animal models (for a review of studies on infant perception and attention, see M. H. Johnson, 1996).
The information-processing approach contains elements of some of the other approaches described previously. Like Piaget, proponents of the information-processing approach believe that perceptual and cognitive development is constructive. According to one view, at least, the emphasis is on infants' learning to process the relationship among properties to form the whole (Cohen, 1988, 1991, 1998; Cohen & Cashon, 2001b). Young infants are able to process simple perceptual properties of objects, such as color, form, and shape, before they can process objects as a whole. As in Piagetian theory, development is also hierarchical. From an information-processing perspective, what counts as a unique whole at one age can serve as a property or element of a larger or more complex whole at an older age. Therefore, after infants perceive or organize perceptual properties into unique objects, they can treat objects themselves as properties of larger wholes and look for relationships between objects in dynamic events involving multiple objects. One such relationship is causality; infants can distinguish causal, direct-launching events from noncausal events. At a later point in development, even these causal events can become elements in more elaborate event sequences.
Recently this approach has been summarized by a set of six propositions (Cohen & Cashon, 2001b):
1. Perceptual-cognitive development follows a set of domain-general information-processing principles.
2. Information in the environment can be processed at a number of different levels of organization.
3. Higher (more holistic) levels can be defined in terms of the types of relationships among lower (parts) levels.
4. Development involves progressing to higher and higher levels.
5. There is a bias to initiate processing at the highest level available.
6. If an information overload occurs (such as when movement is added or when the task involves forming a category), the optimal strategy is to fall back to a lower level of processing.
Much like connectionist modeling and dynamical systems and unlike nativism, the information-processing approach emphasizes the role of experience, learning, and nonlinear changes in development. In fact, attempts are currently underway to produce a connectionist model that follows the preceding six propositions (Chaput & Cohen, 2001). Along with other approaches, visual attention and habituation are frequently used to assess infant information-processing. In contrast to other approaches, it also emphasizes changes in attention and memory over age. An information-processing perspective has often been used in studies of individual differences in preterm versus full-term infants or normal versus aberrant infants as well as long-term predictions of later intellectual ability (e.g., Bornstein & Sigman, 1986; Fagan, 1984).
Was this article helpful?