Form perception in infants is usually studied with two-dimensional or three-dimensional static figures or shapes that have well-defined contours (Slater, 1995b; Slater & Johnson, 1998). The issue most often investigated in form perception is whether infants will respond to the component parts of a shape or to the figure as a whole (Slater, 1995b; Slater & Johnson, 1998). However, making this distinction experimentally is not always easy. For example, Slater, Morison, and Rose (1983) found that newborn infants can discriminate between the outlines of the shapes of a triangle, a square, and a cross. Is this form perception? Perhaps not. In fact, in one of the earliest form perception studies with infants, Salapatek and Kessen (1966) found that when newborns scanned a large triangle, they only scanned a small portion near the apex. To provide clear evidence of form perception, it is important to show that infants are discriminating between these shapes based upon more than just a portion of their outlines or some other component of the figure. It also must be shown that infants process the figure as a whole. As Banks and Salapatek (1983) discussed (see also Slater, 1995b), it is very difficult to obtain unambiguous evidence of form perception because there are often simpler, perceptual explanations for results with infants. Thus, even topics as basic as form perception must deal with issues about part-versus-whole processing that are central to the information-processing approach mentioned earlier.
Fortunately, there are ways to examine this experimental issue. In one such study, Cohen and Younger (1984) investigated developmental differences in the perception of angles by 14- and 6-week-old infants and were able to get clear evidence of form perception in the older infants. Cohen and Younger (1984) tested whether infants would process the parts of the angle—that is, the orientations of the lines—or whether they would process the whole angle—that is, the relationship between the lines. After habituating infants to one angle, they presented variations that either changed the line orientations but not the angle, or that changed the angle but not the line orientations. Their results indicated a developmental shift in the manner in which infants process angles and perhaps other simple forms. The younger infants seem only able to process the line orientations, or the independent parts of the angle, whereas the older infants are able to process the relationship between the lines and process the angle as a whole form.
Slater, Mattock, Brown, and Bremner (1991) conducted a similar set of experiments with newborns. In the first experiment, newborns were found to behave similarly to the younger, 6-week-old infants in the Cohen and Younger (1984) study. Not surprisingly, the newborns in the Slater et al. (1991) study responded to a change in line orientation and not to a change in the angle. In a second experiment, Slater etal. (1991) investigated whether newborns could process the angle independently of its orientation. In this experiment, newborns were familiarized either to an acute or an obtuse angle presented in six different orientations, much like a category study. Infants were then tested on an acute and an obtuse angle, one of which was familiar and the other novel, both in novel orientations. Slater et al. (1991) found that the newborns showed a novelty preference for the novel angle. Slater (1995b) suggested that this could be evidence of form perception in newborns, although he acknowledged that it may not be unambiguous evidence.
One important alternate interpretation has been referred to as the "blob theory" (Slater et al., 1991). This interpretation rests on the notion that at the apex of an angle a low-frequency "blob" is formed, and the size of the blob varies depending on the size of the angle. When newborns discriminated between the acute and obtuse angles in the Slater et al. (1991) test, they may have been responding to the difference in relative size of the blobs and not actually to the angle itself. If this is the case, then the results would be consistent with a developmental progression in the perception of angles, whereby newborns respond to the size of the apex (the blob), 6-week-olds respond to the independent lines of an angle, and 14-week-olds respond to relationship of the lines of the angle or the form of the angle.
Slater and Morison (as cited in Slater, 1995b) also found evidence for a developmental progression in form perception. In this experiment, newborns, 3-month-olds and 5-month-olds were tested on whether they could extract the general shape from a series of figures that varied only slightly from one another in design or texture. After being familiarized to six exemplars of a shape, infants were tested on a novel exemplar of the familiar shape and a novel shape. By showing infants slight variations of the same shape, the experimenters were able to see whether infants could form a category based on the overall form of the shape. If infants were in fact able to form this category, then one would expect infants to show a novelty preference for the novel shape. This is exactly what the 3- and 5-month-old infants did; however, the newborns did not show the preference. These results fit nicely with an information-processing approach and provide further support that form perception may develop over time.
Our knowledge about infants' color perception has grown considerably in the last 25 years (for a review, see Teller & Bornstein, 1987). Before that time, the results of research conducted on infant color perception were somewhat ambiguous. It was never clear in these early experiments whether infants were discriminating between different hues or some other aspect of color, such as brightness or intensity.
In 1975, several researchers invented clever tasks to show that infants younger than 3 months of age can discriminate between stimuli that vary in hue, not just brightness. For example, Peeples and Teller (1975) tested 2-month-old infants on a hue preference test and found that they could discriminate a red hue from a white hue, independent of brightness. More recently, Adams, Courage, and Mercer (1994) tested infants shortly after birth and found that newborns could discriminate red from white, but not blue, green, or yellow from white. As Kellman and Arterberry (1998) concluded, however, by about 2 to 3 months of age, infants seem to have color vision very similar to that of to adults and can discriminate between many colors.
So, within the first 2 to 3 months of life, infants appear to be sensitive to the same spectrum of color as adults. But do infants view the boundaries between colors in the same way as adults? Adults group a range of colors into blue and another range into green, and so on. In other words, do infants (like adults) organize colors into distinct categories? Bornstein, Kessen, and Weiskopf (1976) tested this question with 4-month-old infants. They habituated infants to a stimulus of a certain hue, or wavelength. Then the infants were tested with the same stimulus, a stimulus of a different wavelength but from the same color category, and a stimulus of a different wavelength that was considered by adults to be in a different category. If infants dishabituated to both novel stimuli, one would conclude that they must have responded to the wavelength and not the color categories. However, Bornstein et al. (1976) found that infants dishabituated only to the stimulus that adults considered to be in a different color category. Thus, infants not only perceive colors at an early age, they also seem to organize them into roughly the same color categories as adults.
From an information-processing viewpoint, it is interesting that like form perception, even infants' color perception appears to go through a developmental pattern whereby infants begin by processing information at a lower level, and then later they integrate that information and process it at a higher level. In the case of color perception, infants first gain the ability to discriminate between colors (around 2-3 months of age) and then later, building upon that ability, are able categorize colors (around 4 months of age). In the next section, we examine something that looks very much like categorization with infants' perception of shape and size constancy.
Artists are taught to be conscious of the way they see the world and to create visual illusions such as size, perspective, and distance on the canvas. For example, to create the illusion that an object is farther away, an artist simply draws the object higher and smaller on the page than he or she would draw an object that is meant to be up close. Similarly, adults have little difficulty making sense of the environment and understanding the illusions created on our retina. For example, you would have no trouble recognizing this book as the same book whether you saw it inside on your desk under fluorescent lighting or outside in the bright daylight. You would not be fooled by the different perceptual characteristics of the book due to the different illuminations and would effortlessly understand that it is the same book. Furthermore, you would perceive the book as the same despite its change in location or orientation. This ability to identify an object as the same despite a perceptual transformation is known as perceptual constancy.
One question that researchers have asked is whether young infants see real objects as adults do or retinal images of objects? For example, how would an infant make sense of seeing a teddy bear from across the room and then seeing the same teddy bear up close? Would the infant respond to the objective characteristics of the teddy bear and recognize the two images as the same teddy bear, or would the infant respond to the different-sized images on the retina and perceive the bears as two distinct objects, one much larger than the other? If infants perceived the two images as the same, as adults would, we would say that infants have size constancy—that is, despite the fact that the retinal image of the close teddy bear may be twice as large as the retinal image of the distant teddy bear, the objective size is still preserved.
The notion of constancy can also refer to shape constancy, which is the ability to perceive an object as being the same despite changes to its orientation or slant. For example, if an infant saw a rectangular block from the frontal view and then saw it at a 45° angle, would the infant know that it was the same block? In other words, would the infant (like an adult) understand that despite the change in slant, both objects are the same rectangular block? Or would the infant perceive only the retinal image of these two objects and treat these two as different shapes, one as a rectangle and one as a trapezoid?
Piaget and Inhelder (1969) adhered to the position that infants first respond to the retinal images of objects and believed that infants did not get size constancy until 5 or 6 months of age. They based this belief on the finding that if one taught infants to reach for a large box, the infant would continue to reach for that box even though it projected a smaller image on the retina than did a box that was closer and smaller in real size.
Bower, however, challenged the traditional view of size constancy and was one of the first researchers to test its claims empirically. In several experiments (e.g., Bower, 1966b), he used an operant conditioning paradigm to investigate whether young infants based their responses to an object on that object's real size, retinal size, or distance. Bower (1966b) found that infants generalized their response based upon both the objective size and the object's distance, but not retinal size. Thus, he had evidence that infants younger than 2 months of age do not rely on the retinal size of objects and can respond on the basis of an object's real size and distance.
Day and McKenzie (1981) continued the work on infant size constancy using a habituation paradigm, a completely different technique from Bower's operant conditioning experiments. They also found evidence for size constancy in infants as young as 18 weeks of age. Subsequently, two independent research laboratories tested newborns in a habitua-tion paradigm, and both found evidence of size constancy (Granrud, 1987; Slater, Mattock, & Brown, 1990).
In addition to studying size constancy in infants, Bower also used his operant conditioning technique to study shape constancy. In one experiment with 50- to 60-day-olds, he trained infants on a rectangle that was slanted at a 45° angle, which created a retinal image that looked like a trapezoid.
He then looked for a generalized response to (a) a rectangle at a frontal view (new retinal image, same objective shape, new slant); (b) a trapezoid at a frontal view (same retinal image, new objective shape, new slant); and (c) a trapezoid slanted at a 45° angle (new retinal image, new objective shape, same slant). Bower found that infants generalized their responses to the rectangle presented at a frontal view. This result indicates that the infants responded to the objective shape of objects and not the shape of the retinal image or the slant of the objects.
Caron, Caron, and Carlson (1979) also addressed the issue of shape constancy in young infants, but did so in several studies using a habituation paradigm. Their results supported Bower's finding that young infants perceive the objective shape of objects and do not rely solely on the retinal image of those objects. In fact, in a more recent habituation study, Slater and Morison (1985) also found evidence of shape constancy in newborns.
In sum, the evidence suggests that young infants do not rely solely on the retinal image of objects and are capable from birth (or shortly thereafter) of understanding size and shape constancy. How is it that infants are able to understand these constancies and respond to more than the retinal image of objects at birth? The key may be that all constancies require an understanding or appreciation of relational information. To return to the examples in the beginning of this section, the reason this book outdoors is not perceived as brighter is that relative to other objects, it is not brighter. Furthermore, the reason that an infant would perceive the teddy bear up close and far away as the same bear is that the size is constant relative to the distance of the object. The up close bear may appear two times as large, but it is also two times as close as the distant bear. Thus, the relationship between size and distance has remained the same. It is these constant relationships to which the infants must be sensitive.
Being sensitive to the relationships among things in the world is a necessary requirement for understanding the world around us. From our information-processing perspective, understanding relationships is the central principle around which infant perceptual and cognitive development proceeds. Throughout this chapter, we demonstrate that as infants get older, the types of relationships they process, understand, and remember become more complex and abstract. In that sense, the abilities to understand size, shape, and other constancies become building blocks from which infants learn about first objects and later events in the world about them.
Even some types of constancies may be more cognitively demanding or require more conscious attention to relationships than do others. One of these constancies may be object constancy, which is an understanding that despite a significant physical transformation of an object in space, time, or both, it is the same object. An example would be understanding that the bottle now on its side but seen previously standing up is the same bottle—or recognizing the back of mother's head is the same person who is normally seen from the front. One could go a step further. What if the object or person were not visible at all? For example, consider an infant who hears her mother's voice in the other room and is able to identify the voice as that of his or her mother—or an infant who recognizes that the toy under the table is the same toy he or she had in hand before dropping it and losing sight of it. This extension of object constancy has been examined in great detail in the infant cognition literature by Piaget and many other investigators under the heading of object permanence. Because the development of an understanding about objects has played such a prominent role in investigations of infant cognition, we shall devote considerable space to it in the next section of this chapter.
INFANTS' UNDERSTANDING OF OBJECTS Object Permanence
When people think about what infants know about objects, the concept of object permanence, or understanding that an object continues to exist in the world even though it is hidden or cannot be seen, often comes to mind. There is the general misconception that infants acquire this concept at 8 or 9 months of age. The misconception arises in part because most people think of object permanence as a unitary concept that infants have or do not have at a particular age.
Because one of the most dramatic developments in object permanence, reaching for and obtaining an object that is totally hidden, occurs around 8 or 9 months of age, this is when many assume infants acquire object permanence. However, for Piaget, obtaining a hidden object is only one intermediate step in a long sequence of accomplishments that infants must master during their first 2 years of life (Piaget, 1936/1952, 1937/1954). Obtaining a completely hidden object is characteristic of the onset of Piaget's Stage 4 (9 to 12 months). However, infants at Stage 3 (1 and one half to 4 or 5 months), although not yet able to retrieve a completely hidden object, are able to retrieve an object that is only partially hidden. And even though infants in Stage 4 can retrieve a completely hidden object, if an experimenter subsequently hides the object under a second cloth, infants at this stage will commit what is known as the AnotB error. They will mistakenly go to the first cloth to retrieve the object (for more discussion on the A not B error, see Diamond, 1991; and Haith & Benson, 1998).
Infants at Stage 5 (12-18 months) no longer make this error and will correctly retrieve the hidden object from the correct cloth. However, according to Piaget (1937/1954), Stage 5 infants still do not completely have the concept of object permanence because they are fooled by invisible displacements. If an experimenter shows an infant an object and then places the object in a small box before hiding it under a cloth on the table, the Stage 5 infant will not look for the hidden object at its final destination. An infant who does successfully retrieve an object under this circumstance is considered by Piaget to be in Stage 6 (18-24 months) and to have completely mastered the object concept. (For more discussion on the development of the object concept, see Diamond, 1991; and Haith & Benson, 1998)
More recent research has extended the work of Piaget and considered other questions about infants' understanding of objects. Kellman and Spelke (1983), for example, investigated the role of coordinated movement of an object parts in infants' perception of objectness, or the perception of object unity. In this classic study, they habituated 4-month-old infants to a display in which a partially occluded rod moved back and forth behind an occluder. They were then tested on two displays without the occluder. In one test, infants saw a complete rod that moved back and forth, and in the other they saw just the two rod parts that resembled portions visible during habituation. The infants dishabituated to the two rod parts but not to the solid rod, indicating to Kellman and Spelke (1983) that the parts were novel, and thus that they must have perceived the two moving parts during habituation as a single complete rod. This result, according to Kellman and Spelke, indicates that the infants perceived object unity even under conditions of partial occlusion. It is interesting to note that the perception or inference of an object under conditions of partial occlusion at 4 months of age would be consistent with Piaget's Stage 3 behavior. It therefore would be of interest to test younger infants to see whether this ability to perceive or infer a unified object develops during the first few months of life.
Several researchers have in fact attempted to replicate this object unity study with younger infants. Slater et al. (1990) conducted a similar study with newborns and found a very different set of results from those reported by Kellman and Spelke (1983). Slater et al. found that instead of dishabituating to the two rod parts, newborn infants dishabituated to the complete rod, suggesting that they were perceiving the rod parts as separate items rather than as the top and bottom of a single unified object. More recently, researchers have replicated Kellman and Spelke's findings with 2-month-olds and have found that they dishabituate to the rod parts display in the test if the occluder is rather narrow (S. P. Johnson & Nanez, 1995). Collectively, research on object unity shows that infants are not born with the ability to perceive two parts of a moving, partially occluded object as one object, but that this ability develops over at least the first 4 months of age.
In fact, such a conclusion fits within an information-processing framework by showing once again a developmental change from processing parts to processing wholes. It seems clear from research on object unity and other related topics that well before 7 months of age, infants are capable of perceiving objectness—that is, of perceiving those characteristics that indicate a single unified object exists. From our earlier discussions on form, color, and constancy, it is equally clear that also well before 7 months of age infants perceive many characteristic features of an object. However, as research described in the next section indicates, young infants still lack the ability to distinguish one object from another. This ability has been called object individuation or object segregation and estimates about when it develops range widely from 4 or 5 months of age to 12 months of age.
The ability to distinguish two objects as distinct entities is what researchers refer to as object individuation or object segregation. Depending on the procedure used, there are reports that infants can individuate objects at 12 but not 10 months of age (Xu & Carey, 1996), or in some cases as young as 5 months of age (Needham, 2001; Wilcox, 1999). Xu and Carey (1996) employed an "event-mapping" procedure, as it is referred to by Wilcox and Baillargeon (1998), in which infants were shown an event and then tested on two events—one that was considered consistent and another that was considered inconsistent with the first event. Specifically, infants initially saw one object move behind an occluder from the left, then a different object emerge from behind the occluder and move to the right. The authors reasoned that if infants understood that there were two objects in the original event, then they should look longer at the inconsistent event—that is, the display with one object. They found that the 12-month-olds but not the 10-month-olds, looked longer at the one-object display. Based upon these findings, Xu and Carey concluded that these older infants understood that there were two objects present in the event and had successfully individuated the objects.
However, studies with younger infants suggest that they also—under certain circumstances—can individuate objects. In a very recent set of studies, Needham (2001) gave infants exposure to an object prior to presenting two test events that involved an object similar to the prior-experience-object but varying on some feature such as texture (Experiments 1 and 2), orientation (Experiment 3), and color (Experiments 4, 5, and 6). One test event can be described as two objects move together. In this event, the two objects that are touching and move together could be perceived as either one or two separate objects. The second test event can be described as one object moves. In this event, the object remains stationary, so it should be obvious that there are two distinct objects. Needham reasoned that if infants attended to the featural characteristics of the prior-experience-object and its equivalent in the test was similar enough, then infants should still show signs of individuating the objects in the move-together test event. If, however, infants did not view the two objects as similar, then presumably that prior experience with the first object would not help the infants to individuate the objects in the test display. In this case, infants would not be expected to show a difference in response to the two test events. Needham (2001) found that 4.5-month-olds could use texture and orientation but not color cues to help them individuate the objects in the test displays.
In addition to these findings a study by Wilcox (1999) showed a developmental progression in what featural information can be used to help infants individuate objects. She found that at 4.5 months of age infants can use shape and size, at 7.5 months of age they can use pattern, and at 11.5 months of age they can use color to individuate objects. From our earlier discussions on form and color perception, it is clear that infants in the first 3 months can perceive these characteristics about objects. Furthermore, as the discussion on object unity shows, infants are also capable of perceiving objectness (i.e., that something is a separate object) by 2 months of age. Given this information, one may ask why infants cannot consistently individuate objects until possibly 5 or even 10 to 12 months of age; we believe the answer may lie in an information-processing perspective. The ability to individuate objects requires attention to and integration of both feat-ural information and objectness. Thus, object individuation represents another example of a developmental progression from processing the independent features of objects to integrating or relating those features and processing the object as a whole.
Some theorists believe that infants have sophisticated knowledge about objects and object permanence much earlier than others such as Piaget had assumed. In fact, in many of their studies investigating infants' understanding of objects, an understanding of object permanence is a prerequisite for the infants. For example, in one experiment on young infants' understanding of object solidity—that is, that one solid object cannot pass through another solid object—Spelke, Breinlinger, Macomber, and Jacobson (1992, Experiment 3) showed 2.5-month-olds' events in which a ball rolled from the left end of the stage to behind an occluder. After about 2 s, the occluder was raised to reveal the ball resting against a wall. In the habituation phase, when the occluder was raised, the ball was shown resting on the left side of a wall on the right end of the stage. In the test phase, lifting the occluder revealed two walls, one in the center of the stage and one on the right. In one test event, the ball was found resting against the center wall, which was considered a consistent outcome because the wall would have obstructed the ball's path. In the other test event, the ball rested against the right wall, which was considered an inconsistent outcome because the ball would have had to go through the center wall to reach the right wall.
Spelke et al. (1992) found that infants looked longer at the inconsistent outcome and interpreted this result to mean that infants as young as 2.5 months of age understand that a ball cannot travel through a solid center wall to get to the right wall. However, given the fact that the action of the ball took place behind an occluder, to make the interpretation that these very young infants understand object solidity, one also has to assume that these infants understand the ball continues to exist when behind an occluder—in other words, one must assume that they are operating at least at Stage 4 of object permanence.
Recent evidence with 8-month-old infants and animated events, however, suggests a simpler explanation for this apparent sophisticated cognitive ability of infants. It is possible the infants were simply responding to changes in the perceptual cues of the events, such as the duration of movement or the presence of a wall to the left of the ball (Bradley & Cohen, 1994). In another, similar set of experiments on infants' understanding of solidity, Cohen, Gilbert, and Brown (1996) tested 4-, 8-, and 10-month-old infants. They found that infants had to be at least 10 months of age before they really understood that one solid object cannot pass through another solid object.
If this conclusion is accurate, once again it fits within the information-processing framework. As we have mentioned, there is evidence that infants first learn about the independent features of objects by about 4 months and then integrate these features into a whole object by about 7 months. The next developmental step would be for infants to understand the relationship between objects—in this case, to understand that one solid object cannot pass through another solid object. It makes sense to us that the ability to understand object solidity may not develop until approximately 10 months of age, given that infants would first have to be able to individuate and segregate individual objects.
Another type of relationship between objects would be a causal relationship, the simplest version of which would occur in what has been called a direct launching event. In this type of event, one moving object hits a second moving object, causing the second object to move. Several studies have now been reported on infants' perception or understanding of causality in these types of events. Once again, as predicted by an information-processing view, when realistic objects are being used, infants do not perceive the causality until approximately 10 months of age. See Cohen, Amsel, Redford, and Casasola (1998) for a review of this literature.
In addition to Spelke and her colleagues, Baillargeon has also reported a large body of research suggesting that infants are precocious (for reviews, see Baillargeon, 1995, 1999). In one of her most well-known set of studies, she reported that infants as young as 3.5 months old understand that an object continues to exist when it is out of sight (Baillargeon, 1987; Baillargeon et al., 1985). Her procedure was very different from the traditional Piagetian object permanence task. Instead of relying on an infants' ability to reach for a hidden object, she (like Spelke) relied on infants' looking times at possible and impossible events.
The procedure involved familiarizing infants with a screen that rotated 180° back and forth, from a position of lying flat on the front part of a stage to lying flat on the back part of the stage. Infants were then tested on a possible and an impossible event, both of which involved the presence of an object such as a yellow box placed in the path of the rotating screen. In the possible event, infants saw the box resting on the stage before the start of the first rotation. The screen then rotated back and forth, as it did in the familiarization event. Each time it rotated back, it hid the object from the infant. Furthermore, the screen stopped rotating at 112° when it appeared to make contact with the object and then rotated back toward the infant, once again reexposing the object. The impossible event was similar to the possible event, except that the screen rotated a full 180°, appearing to go magically through the space that should have been occupied by the box. (There was actually an experimenter behind the stage who removed the box so that the screen could complete its rotation. As the screen rotated toward the infant again, the experimenter replaced the box on the stage in time for the infant to see the box once again resting in the screen's path.)
Baillargeon (1987) found that 4.5- and some 3.5-month-old infants looked longer at the impossible event. She interpreted this finding to mean that the infants "understood that (a) the object behind the screen continued to exist after the screen rotated upward and occluded it and (b) the screen could not move through the space occupied by the object" (p. 662). She based these interpretations on several assumptions:
(a) Infants normally have a novelty preference during the test phase, (b) infants would perceive the impossible event as familiar because the amount of rotation in this event is the same as the amount of rotation in the familiarization event, (c) infants would perceive the possible event as novel because the amount of rotation is novel, and (d) if infants looked longer at the impossible event, which should be perceived as familiar, it must be for reasons other than novelty; it must be because they understood object permanence and object solidity and were observing a violation of both concepts.
However, as with Spelke et al. (1992), recent evidence suggests a simpler, perceptual explanation of Baillargeon's so-called drawbridge results (Baillargeon, 1987; Baillargeon et al., 1985). We have already mentioned the problem of a familiarity effect in habituation-familiarization studies. Along those lines, one alternative interpretation is that the infants in Baillargeon's studies were not fully habituated and thus did not have a novelty preference during the test phase, as she assumed. The results of these more recent studies, which varied familiarization time or used more stringent habituation criteria, support the interpretation that infants looked longer at the impossible event because it was familiar, not because it was impossible (Bogartz, Shinskey, & Schilling, 2000; Cashon & Cohen, 2000; Schilling, 2000; see also Bogartz, Cashon, Cohen, Schilling, & Shinskey, 2000).
The findings that 3.5- to 4.5-month-olds understand object permanence (Baillargeon, 1987) and that infants as young as 2.5 months old understand object solidity (Spelke et al., 1992) stand in stark contrast to Piaget's reported ages and stages. Thus, if one assumes that explanations like Baillargeon's (1987) and Spelke et al.'s (1992) are correct regarding infants' early understanding of object permanence, that explanation has to be reconciled with the fact that under standard object permanence techniques, infants do not show evidence that they understand an object exists when completely hidden until at least 8 or 9 months of age.
The prevailing explanation for this discrepancy is that younger infants understand that hidden objects continue to exist, but they fail to reach for those objects in a standard Piagetian task because they have difficulty with means-end actions (Baillargeon, 1987; Baillargeon et al., 1985; Bower, 1974; Diamond, 1991). In other words, infants may have trouble coordinating two actions to obtain a goal—in this case, removing the cloth and then reaching for the object. Once again, however, recent evidence suggests that infants do not have a means-end deficit. A couple of recent studies, for example, have shown that infants do not have the same reaching problem when the object is behind a transparent obstacle versus an opaque obstacle (Munakata et al., 1997; Shinskey, Bogartz, & Poirer, 2000). Taken together, these more recent results uphold previous findings that infants younger than 8 or 9 months of age fail to search for hidden objects not because they lack a means-end skill, but because they have yet to understand that objects continue to exist when they are hidden.
A considerable amount of research has been conducted on infants' perception of faces over the past 40 or so years (see Maurer, 1985, for review). There is no question that faces are important stimuli for infants. Infants see faces often and use them to help identify others, interact with others, and learn about the world. It may seem odd to some that we have included a section of face perception within a section devoted to objects. However, one issue that arises in the study of infants' perception of faces is whether infants view faces as something special or whether they perceive faces in the same way they perceive other complex objects (see Kleiner, 1993, for discussion). Nativists often argue that faces are a unique class of objects and that the way in which newborns process a face is quite different from the way in which newborns process nonface stimuli (e.g., Fantz, 1961; Morton & Johnson, 1991). Empiricists, however, regard face perception quite differently. They argue that the way in which we process a face is brought about through experience; at least in the beginning, faces are no different from other objects (e.g., Banks & Ginsburg, 1985). In this section, we review some of the literature regarding the issue of infants' preference for facelike over nonfacelike stimuli, followed by a discussion of how infants process faces and how that processing may change with age.
Research on infants' perception of faces has produced conflicting results with respect to the question of whether faces are special to infants. Whether newborns have an innate preference to look at faces over other stimuli is still unresolved (for discussions, see Easterbrooks, Kisilevsky, Hains, & Muir, 1999; Maurer, 1985). Visual tracking studies with newborns have shown that neonates will follow (with their eyes) a facelike pattern farther than they will follow a nonfacelike pattern (Goren, Sarty, & Wu, 1975; M. H. Johnson, Dziurawiec, Ellis, & Morton, 1991; Maurer & Young, 1983). However, preferential looking paradigm studies have provided a different picture. Fantz and Nevis (1967) found that in general, newborns preferred to look at patterned stimuli, such as a bull's-eye and a schematic face, to plain stimuli, and preferred a schematic face to a bull's-eye; however, they did not find a preference for a schematic face over a scrambled face. In another preferential looking study, Maurer and Barrera (1981) found a preference for a facelike pattern over scrambled faces at 2 months of age, but not at 1 month of age. M. H. Johnson et al. (1991, Experiment 2) replicated Maurer and Barrera's findings with 5- and 10-week-old infants in a preferential looking paradigm and Goren et al.'s (1975) with newborns in a visual tracking task.
It may seem odd that the evidence suggests that newborns have a preference for faces but that this preference disappears by 2 months of age. One possible explanation for the discrepancy, posited by Morton and Johnson (1991) and Johnson and Morton (1991), is that two different testing methods were used that may tap into two different processes of face recognition in place at different ages. The first mechanism they describe is CONSPEC, a subcortical device in the brain of the newborn that contains the information about the structure of a face. CONSPEC is believed to attract infants' attention to stimuli with the same structural information as faces, which would account for newborns' preferential tracking of faces. The second mechanism, CONLERN, is thought to take over by the second month. It is assumed to be a cortical structure that is involved in learning about conspecifics of a face. This mechanism is believed to help in the recognition of individual faces.
If Morton and Johnson are correct that infants have an innate representation of the structure of faces, then we could certainly conclude that faces are special. However, evidence from other studies raised doubts about this conclusion. The results of studies on infants' visual scanning patterns of faces and nonface stimuli have revealed similar scanning patterns and developmental trends for faces and nonface stimuli alike, which suggests that faces may not be special to infants (see Salapatek, 1975, for discussion). Several researchers have reported finding that infants tend to scan mostly the external contour of a face in the first month of life, whereas in the second month, infants tend to scan the internal features (Bergman, Haith, & Mann, 1971, as cited in Salapatek, 1975; Maurer & Salapatek, 1976; Salapatek, 1975). This developmental shift in the scanning pattern of faces has also been found in infants' scanning pattern of nonface stimuli (Salapatek, 1975).
A number of investigators have examined the development of infants' face processing over the first year of life and whether that processing is similar to or different from the development of object processing. Previous findings with 4- and 7-month-olds have shown that the younger infants process the independent features of line-drawn animals, but that the older infants are sensitive to the correlations among features (Younger & Cohen, 1986). More recently, we have been investigating whether this developmental shift from parts to whole processing would also be true in 4- and 7-month-old infants' processing of faces as well. It has been found that adults process upright faces as a whole but inverted faces in a piecemeal manner. Therefore, we also examined the effect of orientation. Half of the infants saw all upright faces and the other half saw inverted faces.
Some of the results were expected; others were surprising. The 7-month-olds behaved as expected—that is, they responded to upright faces as a whole but inverted faces in a piecemeal fashion (Cohen & Cashon, 2001a). The 4-month-olds, however, behaved in a totally unexpected manner. In one sense, they appeared to be more advanced than the 7-month-olds—or even adults. Not only did these younger infants process a face as a whole when it was presented upright, but they also processed it as a whole when it was presented in an inverted orientation (Cashon & Cohen, 2001)! One possible explanation for this finding currently under investigation is that an upright facial orientation may not be as important to infants at 4 months of age. They undoubtedly receive a considerable amount of exposure to faces in a variety of orientations, perhaps much more so than at 7 months when they are stronger and tend to view the world from an upright position.
Was this article helpful?