INVITED ARTICLE
NATIVISM, EMPIRICISM, AND THE ORIGINS OF KNOWlEDGE
Elizabeth S. Spelke MIT
What aspects of knowledge emerge in children Although contributors have tended to be prior to their first contacts with the objects of labeled “nativists” or “empiricists” according their knowledge, and what aspects emerge to the kinds of answers they thought most plau- through the shaping effects of experience with sible, most have viewed these questions as those objects? What aspects of knowledge are empirical matters to be resolved not by ideol- constant over human development from the ogy but by studies of the origins and develop- moment that infants begin to make sense of the ment of knowledge. Research on cognition in world, and what aspects change as children infancy remained a dormant enterprise grow and learn? What aspects of knowledge throughout most of the history of the nativ- are universal, and what aspects vary across ist-empiricist dialogue, however, because the people in different cultures or with different tools then used to probe human knowledge educational backgrounds? Finally, what were not appropriate for young children. aspects of knowledge can people change in Today, the study of early cognitive devel- themselves or their children with sufficient opment has overcome this longstanding barrier insight or effort, and what aspects are invari- to progress. A number of tools have been ant? developed over this century for investigating These questions are central to a dialogue human cognitive states and processes, and that has spanned more than 2000 years of intel- some of these tools have been adapted for stud- lectual history. Contributors to the dialogue ies of preverbal children. New tools of enor- have raised the questions in order to shed light mous promise are appearing, moreover, with on larger concerns about human nature, child the rapid development of cognitive neuro- development, education, science, and society. science. For the first time, these tools allow
Editor’s Note: The Haith and Spelke articles were presented in a debate at the meeting of the Society for Research in Child Development, Washington, D.C. April 1997.
-Elizabeth S. Spelke, EIO-246, MIT, Cambridge, MA 02139; e-mail: [email protected].
INFANT BEHAVIOR & DEVELOPMENT 21 (2), 1998, pp. 181-200 ISSN 0163-6383 Copyright 0 1998 Ablex Publishing Corporation All rights of reproduction in any form reserved. 182 INFANT I3EHAVIOK R DEVELOPMENT Vol. 21. No. 2, I’IOti developmental scientists to use studies of OBJECT REPRESENTATION IN infancy to shed light on the central questions of INFANCY the nativist-empiricist dialogue. As ancient obstacles have been overcome, Human adults perceive their surroundings as a however. new obstacles have arisen. Counter- layout of continuous surfaces furnished with ing the advance of research are intellectual material objects. These objects typically are attitudes that impede studies of cognition in represented as internally connected and exter- infancy and undermine the larger questions nally bounded, with surfaces that continue those studies address. Investigations of infant behind nearer, occluding objects. When an cognition are sometimes dismissed on the object moves, it is represented as behaving in a grounds that young infants are known LIpriori coherent manner. and this representation sup- to be incapable of true knowledge or cognitive ports predictions about the object’s future processes. and investigator\ are sometimes behavior. When motion carries an object fully handicapped by demands that no empirical out of view. the object continues to be repre- enterprise can meet. The questions of the sented: such representations guide actions on nativist-empiricist dialogue have lost much 01 hidden objects. their allure, moreover, because of widespread For centuries, contributors to the nati\- arguments that claims for inrlate knowledge ist-empiricist dialogue have puzzled over the are incoherent, false, or dangerous to society. origins and development of these abilities. and In this article, 1 argue that our intellectual a spectrum of possibilities have been envis- ancestors were right to ask the questions of the aged (Figure I ). At one extreme, object repre- nativist-empiricist dialogue, and that develop- sentations might bc shaped entirely by mental scientists should address these ques- children’s perceptual encounters with ob.jects. tions vigorously through research on early All abilities to represent objects as bounded. cognitive development. My defense of the dia- persisting bodies with predictable motions logue is divided into three parts. First, 1 discuss might arise as children explore ob.jrcts and dis- one example of research on cognition in cover that they have these properties. At the infancy-studies of object representation-in opposite extreme, object representations might hopes of showing how this research is advanc- emerge entirely by virtue of intrinsic processes ing understanding of the origins and develop- of growth, independently of any specific ment of knowledge. Second, I consider some encounters with objects. Between these contemporary critiques of this research. Argu- extreme\ ure a wealth of intermediate po\sibil- ing that the criticisms are based on skewed ities, because certain aspects of mature okject interpretations and impossible standards. 1 representation\ may stem from intrinsic prop suggest a different set of standards against erties of humans’ perceptual and cognitive ~I’x- which all research on early cognitive develop- tems whereas other aspects may stem from ment could productively be evaluated. Third, I learning about the particular characteristics of consider some popular, contemporary argu- surrounding objects. ments against the nativist-empiricist dialogue, Although many contemporary investiga- focusing in particular on arguments against tors appear to regard the firxt. radically any claim that knowledge can emerge through empiricist hypothesis as mo\t plausible (e.g.. intrinsic growth processes. without prior shap- Baillargeon. I99.3; Elman, Bates. Johnson. ing by encounters with the objects of knowl- Kurmiloff-Smith, Parisi, bli Plunkett. I996; edge. I conclude that the arguments are Haith. 1997: Munakata. McClelland. Johnson. mistaken and that the concerns that motivated & Sicgler. 1997: Thelen & Smith, 1994). thih them instead should lead developmental scien- predilection is not supported by evidence tists to embrace the dialogue and pursue from studies of ob.ject representation in research on the origins of knowledge. infancy. In my view. such stud& have not Ndtivkm, Empiricitm, dnd Knowledge 183
wholly wholly shaped by independent encounters of encounters with objects with objects
FIGURE 1 Theories of the development of object representations within the nativist-empiricist spectrum.
yet eliminated any region of the spectrum in briefly the findings of many studies,’ there is Figure 1, and so investigators need to con- evidence that infants are capable of forming sider the entire spectrum of possibilities. For- certain object representations before they can tunately, studies of infants have greatly act on objects effectively, and also evidence reduced the density of tenable developmental that object representations undergo changes hypotheses within this spectrum, bringing over the time period when reaching and the questions of the nativist-empiricist dia- manipulation develop. logue into greater focus. I believe these stud- Consider, for example, infants’ representa- ies also suggest that intermediate positions in tion of the boundaries of objects in visible Figure I are more plausible than positions at scenes. Perception of object boundaries has either extreme. been investigated by preferential looking meth- ods, focused on infants’ novelty reactions (longer looking) to arrays in which the bound- Object Representations in aries of objects are changed, and also by reach- 3-6 Month Old Infants ing methods, focusing on infants’ tendency to Over last 25 years, many studies of the direct their hands toward the perceived edges of early development of object representation objects. Converging conclusions emerge from have focused on infants in the second trimes- these two lines of research: even the youngest ter of postnatal life. The time from 3 to 6 infants tested perceive object boundaries in cer- months is of theoretical interest, because most tain visible scenes, but their perceptions are empiricist theories have rooted the develop- considerably less specific than those of adults. ment of object representations in actions such By 3-4 months of age, infants perceive fig- as reaching for objects, manipulating objects, ure-ground relationships by analyzing the rela- and moving through the spatial layout (e.g., tive motions and depth relations among visible Berkeley. 1709/l 975; Helmholtz, 18671 surfaces , but not by analyzing the two-dimen- 1962).’ Between 3 and 6 months of age, most sional Gestalt relations among surfaces (Ter- infants begin to engage in object-directed mine, Hrynick, Kestenbaum, Gleitman, & reaching and manipulation, and some infants Spelke, 1987; Yonas & Granrud, 1985). Simi- begin to locomote independently. Investiga- larly, 3-month-old infants perceive the bound- tions focused on these ages therefore can dis- aries of adjacent objects when the objects are cover whether any object representations separated in depth or undergo different emerge before the onset of these activities, motions. When the objects are adjacent and sta- and how object representations change once tionary, young infants are less sensitive to their these activities have begun. To summarize boundaries (Hofsten & Spelke, 1985; Spelke, 1114 INFANT HEHAVIOK & DEVELOPMENT Vol. 21, No. 2, 1998
Hofsten. & Kestenbaum, 1989; Spelke, Brein- 199 1: Craton & Yonas, 1990; Hood & Willats, linger, Jacobson, &Phillips. 1993; Xu & Carey. 1986; Rochat & Hespos, 1996; Wynn, 1992). 1996; although see Needham, Baillargeon, & Under certain conditions, infants also repre- Kaufman, 1997). sent fully occluded objects whose separate Infants’ perception of the continuity of an parts move into view at different times (Van de object behind a partial occluder has been Walle & Spelke, 1996). although this ability investigated primarily with preferential look- shows striking limits (Arterberry, 1993). ing methods, in which infants are familiarized Infants are capable of representing at least two with a center-occluded object and then then hidden objects within a single scene (Baillar- novelty reactions are observed to displays in geon, 1986; Wynn, 1992) but their representa- which the occluder is removed to reveal either tions appear to break down when large1 a gap (novel for adults) or a continuous object. numbers of objects are occluded (see Chiang These studies provide evidence that 4- & Wynn, 1996). Infants also can extrapolate month-old infants perceive a center-occluded the motions of occluded objects in accord with object as continuous when its visible surfaces certain constraints on object motion: for exam- undergo common motion (Johnson & Aslin, ple. they infer that interacting inanimate 1996: Johnson & Nafiez, 1995; Jusczyk, objects change their motions on contact (Ball, Johnson, Spelke, Kennedy, & Smith. 1997; 1973; Leslie & Keeble, 1987). Young infants Kellman & Spelke, 1983; Slater, Morison, are not sensitive to all the constraints on Somers, Mattock. Brown. & Taylor. 1990b). objects that adults recognize. however, and Perception of the continuity of a partly they do not represent object properties as occluded object also is affected by the align- robustly as do adults (e.g., Spelke, Katz, Pur- ment relations among the object’s visible sur- cell, Ehrlich, & Breinlinger, 1994; Xu & faces (Johnson & Aslin, 1996), although adults Carey, 1996). show a greater effect of edge alignment than The findings of all the above studies pro- do infants (Kellman & Spelke, 1983). Finally, vide evidence for an early-developing system infants’ perception of center-occluded objects of object representation that operates in accord does not appear to be affected by either syn- with general constraints on object motion chronous changes in a stationary object’s brightness or hue (Jusczyk, et al.. 1997) or by (Leslie, 1994; Spelke and Van de Walle, 1993) differences in the color and texture of a station- and, to a lesser degree, in accord with Gestalt ary object’s visible surfaces (Kellman & relationships such as edge alignment (Johnson Spelke, 1983; cf. Needham, 1994).’ & Aslin, 1996; Needham, 1997; van Giffen & Further experiments have investigated Haith. 1984). In the context of the nativ- infants’ representation of the continuing exist- ist-empiricist dialogue, we may ask how this ence of objects that are fully occluded. Some system develops. Because 3-month-old infants experiments have used preferential looking do not yet reach for objects or crawl around methods. focusing on infants’ novelty reac- them, knowledge of basic properties of objects tions to events in which visible object motions does not emerge through shaping effects of violate physical constraints imposed by the these actions. Instead, early-developing object existence and location of hidden objects; other representations likely emerge either through experiments have used reaching methods, prior visual experience with objects or through focusing on infants’ reaching for objects in intrinsic growth processes. Studies of younger darkness. These studies provide evidence that infants have not yet distinguished these possi- 3- to 6-month-old infants represent the contin- bilities. but they are progressing toward that uous existence of an object that is first visible goal. To illustrate, I discuss one recent line of and then fully occluded (e.g., Baillargeon, research focusing on very young infants’ per- 1993; Clifton, Rochat, Litovsky. & Perris, ception of partly occluded objects. Nativism, Empiricism, and Knowledge 185
Representation of Partly Occluded object perception result from changes in depth Objects from Birth to 4 Months perception, motion perception, or object per- ception proper. Although existing research Although newborn infants do not reach for does not fully resolve this question, investiga- objects, they show systematic looking prefer- tors are very close to an answer. ences (Fantz, 1961) including a preference for Slater, Johnson, Kellman and Spelke (1994) novel displays over familiar ones (e.g., Fried- investigated whether developmental changes man, 1972). Investigators therefore have used in object perception resulted from changes in variants of Kellman and Spelke’s (1983) pref- depth perception by presenting newborn erential looking method to investigate whether infants with occluded object displays contain- very young infants perceive a center-occluded ing enhanced depth information known to be object to continue behind its occluder. In the detectable at that age (Slater, Mattock, & first study using this method, Slater, et al. Brown, 1990a). Infants’ looking preferences (1990b) confirmed that 4-month-old infants were not affected by this manipulation, sug- perceive the unity of a moving, center gesting that developmental changes in sensi- occluded object but found that newborn infants tivity to depth do not account for do not: After familiarization with a cen- developmental changes in perception of partly ter-occluded object, newborn infants looked occluded objects. Johnson and Aslin (1995) longer at a connected object than at a display next investigated whether developmental with a gap where the occluder had been. This changes in object perception resulted from looking preference, opposite to that of the changes in sensitivity to motion relationships older infants, suggests that newborn infants within a visible scene. They presented fail to perceive the unity of a moving, center 2-month-old infants with partly occluded occluded object. Subsequent research by object displays in which the detectability of Johnson and Naiiez (1995) revealed a transi- common motion was enhanced through three tional looking pattern at 2 months of age: After separate, ingenious manipulations of the familiarization with a center-occluded object occlusion display (see Johnson & Aslin, 1995). similar to those used by Kellman and Slater, Under all three conditions, 2-month-old 2-month-old infants showed no preference infants succeeded in perceiving the unity of a between a complete object and an object dis- center-occluded object. This finding provides play with a gap. These studies provide evi- evidence that 2-month-old infants have a func- dence for a developmental change in tional system for representing partly occluded perception of the visible surfaces of Kellman’s objects, and that limits on motion sensitivity center-occluded objects, from unconnected account for their failure to perceive such (newborn) to ambiguous (2 months) to con- objects when tested with Kellman’s original nected (4 months). displays. Newborn infants also may have a The discovery of this developmental functional system of object representation. but change allows investigators to pose more existing experiments do not address this poasi- focused questions: What perceptual capacities bility. are developing over this period, and what causes their development? Successful repre- Object Representations in Newborn sentation of the unity of a moving, cen- Chicks ter-occluded object requires that infants perceive the three-dimensional arrangements Newborn human infants’ poor acuity and and motions of surfaces in the visible layout motion sensitivity may mask a number of per- and then group these perceived surfaces into ceptual abilities at the start of postnatal life. objects (Figure 2). One may investigate, there- complicating the task of students of perceptual fore, whether the developmental changes in development (see Banks & Shannon. IW3 ). INFANT BEHAVIOR & DEVELOPMENT Vol. 21, No. 2, 1991)
detection of surface depth organization of surfaces into objects detection f of surface motion
FIGUKL2 Processes underlying perception of moving, center-occluded objects at 4 months of age (after Kellman & Spelke, 1983). Because 4 month-old infants perceive partly hidden objects as connected only when their surfaces undergo common motion and stand behind a nearer occluding object, failure to perceive partly hidden objects at younger ages could stem from limits to any of the component processes.
Fortunately, comparative studies of the anat- “imprinting” in a novel test chamber by main- omy, physiology, and functional organization taining proximity to the familiar object. of the visual system suggest that many of the located at one end of the chamber, relative to a basic perceptual mechanisms found in humans novel object located at the other end of the are shared by other vertebrates. If mechanisms chamber. The investigators therefore used this for perceiving and representing objects are not measure of imprinting to assess the chicks’ unique to humans, then insights into the early representations of occluded objects. In a series development of object representations may of studies, chicks who were imprinted to the come from studies of other animals whose sen- fully visible triangle were presented with sory systems are more mature at birth. occlusion displays for the imprinting test (see Recent studies have investigated object rep- Figure 3). Although the chicks had never seen resentations in newborn chicks, using an any occlusion display before, they showed imprinting method (Lean, Slater & Regolin, imprinting to a center-occluded triangle, rela- 1996; Regolin & Vallortigara, 1995). Regolin tive to non-occluded displays containing the and Vallortigara placed chicks in a cage con- same visible surfaces of the triangle. This find- taining a single visible triangle. Because the ing and others (see Lea et al., 1996; Regolin & object was dangled from the center of an other- Vallortigara, 199.5) provide evidence that wise empty chamber, a chick never saw the chicks. like 4-month-old human infants, per- object occluded by any other object. After two ceive center-occluded objects as connected. days’ exposure to the object, chicks exhibited Mechanisms for representing the complete Nativism, Empiricism, dnd Knowledge
Imprinting
Test
FIGURE 3 Displays and apparatus for studies of perception of center-occluded objects by Z-day-old chicks. After spending their first days in a cage with a triangle (top), chicks are tested with occluded and interrupted tri- angle displays (bottom) (after Regolin & Vallortigara, 1995). shapes of partly hidden objects evidently are striking ways from those of adults, providing innate in chicks, for they are present and func- evidence for developmental changes in some tional the first time a chick sees an occlusion perceptual or cognitive mechanisms. For display. As research by Slater, Johnson, Aslin decades, investigators have tried to understand and other investigators continues, we may these changes through further studies of object learn whether these abilities are innate in representation in infants. Here, I focus on one human infants as well. much-studied change, reflected in children’s developing abilities to search for fully Development of Object occluded objects. The landmark research of Piaget (1954) Representations after 6 Months revealed striking limits in young infants’ As noted earlier, the object representations search for occluded objects. Until about 9 of 3- to 6-month-old infants differ in some months of age, infants do not attempt to reach 188 INFANT IjEHAVIOK & DEVELOPMENT Vol. 21, No. L, lW8
for objects that are visibly occluded, even pressing a button. After training, infants though younger infants reach for objects received a succession of trials in which either obscured by darkness (Hood & Willats, 1986; the object or the empty stage was covered by Clifton et al, 199 I ) and give evidence of repre- an opaque or transparent occluder. Infants senting visibly occluded objects in their look- acted to retrieve the object when it was present ing behavior (e.g., Ahmed & Ruffman, 1996; more than when it was absent in the conditions Baillargeon, 1993). Starting at about 9 months, with the transparent occluder (indicating that infants begin to obtain occluded objects by they were capable of performing a differenti- reaching around or displacing their occluders. ated search response) but showed no such dif- What accounts for this developmental change? ference with the opaque occluder. Second, According to one family of hypotheses, Hofsten. Spelke, Feng, & Vishton (1994) developmental changes in search for occluded investigated infants’ reaching for a moving objects stem from changes in the infant’s object that entered reaching space after either action capacities. Successful search may moving on a continuously visible path or mov- depend on emerging abilities to coordinate ing briefly behind an out-of-reach occluder. actions into means-ends relationships (Piaget, Although the object could be obtained by a 1952), to inhibit prepotent actions on visible simple, direct reach under both these condi- arrays (Diamond, 1990a; Thelen, 199S), or to tions, infants’ reaching was greatly perturbed reach on indirect paths (Diamond, 1990b; by the occluder. In both situations, infants Noland, 1996). Studies of chicks-a species failed to engage in actions within their reper- with more precocial behavioral as well as per- toire that would have sufficed to obtain a ten- ceptual capacities-are consistent with these porarily occluded object. Limits on accounts. for newborn chicks have been found sensory-motor coordination therefore are not to solve “object permanence” tasks failed by sufficient to account for infants’ search fail- &month-old human infants (Regolin, Vallorti- ures. gara & Zanforlin, 1995). Developmental These findings suggest limits to infants’ changes in human infants’ action capacities representations of occluded objects, but what therefore may contribute substantially to are the nature and sources of these limits‘? developmental changes in search for occluded objects. But are they the only source of Although research has not fully answered this changes in children’s reactions to hidden question, suggestions come from a recent objects? experiment by Munakata (1997). Munakata The hypothesis that object representations hypothesized that 6-month-old infants repre- are invariant over the development of object sent both occluded and visible objects, and search, and that only changes in action capaci- that their representations have two properties ties produce the dramatic changes in infants’ found also in adults. First, representations of behavior, leads to a straightforward prediction: visible objects are stronger than representa- All developmental changes in object search tions of occluded objects: Objects are experi- should disappear when infants are given search enced more vividly and in greater detail when tasks that do not require means-ends coordina- they are directly visible. Second, representa- tion, suppression of prepotent responses, or tions of different. simultaneously present indirect reaches. Two lines of experiments objects compete with one another for atten- have tested this prediction and disconfirmed it: tion: As the number of objects in a scene Six-month-old infants fail to retrieve occluded increases, the amount of attention devoted to objects even when all the above action any one object declines. Putting these two demands are minimized or eliminated. properties together, Munakata hypothesized First, Munakata, et al. ( 1997) trained infants that competition from a strong representation to retrieve an object by pulling a blanket or of a visible occluder weakens (but does not Nativism, Empiricism, and Know/edge 189 fully abolish) infants’ concurrent representa- through the voluntary direction of attention. If tion of an occluded object. this suggestion is correct, then there is both Munakata’s thesis led to an otherwise coun- constancy and change in object representations terintuitive prediction: When an infant views a over early cognitive development, explaining moving object that is briefly occluded before both infants’ early-developing capacities and entering reaching space, the suppressive effect some of the limits on those capacities. of the occluder will diminish, and reaching In summary, basic questions about the ori- will increase, if the infant is plunged into dark- gins and early development of object represen- ness: Introduction of a blackout period at the tations are still outstanding, but progress is time of occlusion will enhance reaching for the being made. The most popular developmental occluded object. This prediction has received theories of past generations can now be an initial test in a predictive reaching experi- rejected, and the set of tenable theories, ment in which an object’s visibility was briefly although still large, has been narrowed signifi- interrupted by occlusion, by a blackout period, cantly. Most important, recent research sug- or by both occlusion and blackout. Although gests that investigators have the tools to make 6-month-old infants showed low levels of further progress by continuing current research reaching in all 3 conditions, reaching levels trajectories. These tools are being supple- were higher when the loss of visibility was mented by new methods from cognitive neuro- caused by the blackout period than when it was science (e.g., Casey, et al, 1997; Dehaene- caused by occlusion, consistent with the find- Lambertz & Dehaene, 1994) and they are ings of previous studies (Hood & Willats, being extended to probe early cognitive devel- 1986). Most important, the combination of opment in other domains of knowledge includ- occlusion and blackout led to levels of reach- ing knowledge of number (e.g., Wynn, 1995), ing as high as that for blackout alone and of object categories (e.g., Mandler & McDon- higher than that for occlusion alone. Munakata ough, 1993), and of people (e.g., Gergely, concluded that the blackout period diminished Nadasdy, Csibra, & Biro, 1995; Woodward, the strength of the representation of the 1995; Wu, 1997). All these investigations, occluder and thereby strengthened the repre- however, face a serious impediment. sentation of the occluded object. This new finding suggests an explanation for part of the developmental change in search CHALLENGES TO THE STUDY OF for occluded objects. At all ages, including COGNITION IN INFANCY early infancy, humans may be capable of rep- resenting occluded objects. At all ages, more- Like all empirical research, studies of cogni- over, representations of visible objects may be tion in infancy can thrive only in an environ- stronger than representations of hidden ment in which investigators are open to any objects, and representations of distinct objects discoveries their research might yield, includ- within a single scene may compete for atten- ing evidence for knowledge in the mind that tion. These properties of object representations did not arise through the shaping effects of may combine to make actions on occluded sensory contact with the things that are known. objects more difficult, at all ages, than actions A number of students of development are per- on visible objects. With development, how- suaded, however, that such openness is inap- ever, children may become increasingly adept propriate, and that the field should reject either at deploying attention so as to boost activation the questions at the center of the nativ- of particular object representations in relevant ist-empiricist dialogue or any answer to those task contexts. The boost in object representa- questions short of extreme empiricism. When tions that young babies get from a period of minds are closed, research can only suffer. blackout may come to older infants and adults Here, I consider a family of skeptical attitudes 100 INFANT BEHAVIOR It DEVELOPMENT Vol. 21, No. 2. 1 WH
to research on infant cognition, first discussing off every possible perceptual interpretation of skeptical reactions to specific research find- differences [in looking times] to entertain ings and then discussing the prevalent attitude default cognitive interpretations.” That is, no of wariness toward nativist claims. evidence for any cognitive ability in infants can be accepted until every sensory and per- ceptual interpretation of the evidence, ho~rl~r- Standards for Research on impluusihle md rmpiricully unsupported. has Cognition in Infancy been eliminated. Anyone who has conducted research on For example, Haith (1997) considered perception or cognition in infants has likely Wynn’s (1992, 1995) experiments, in which encountered colleagues, science writers, and the looking patterns of infants who viewed a others who have expressed disbelief at his or succession of occlusion events on a single her findings. Evidence for perceptual and cog- stage provide evidence that the infants repre- nitive capacities in infants strains the beliefs of sented two objects on the stage, even though many people because it conflicts with preva- only one object was visible at a time. He lent conceptions about infants and intuitions argued that infants’ looking patterns should about cognitive development. Haith ( 1997) not be interpreted as evidence for object repre- states this conflict clearly and casts his lot on sentations. because there is an alternative the side of intuition, criticizing students of interpretation that has not been eliminated: infant cognition for “asserting that young infants’ looking patterns could be produced by infants know things about objects, events and extremely long-term sensory persistence people far earlier than seems reasonable.” evoked by each object before it was occluded. When data conflict with intuition, however, Haith’s alternative interpretation is implausi- intuition is rarely the best guide for advancing ble, because abundant research with adults understanding. Intuition has proved to be an provides evidence that sensory persistence in especially poor guide to understanding human lighted environments is at least an order of perception and cognition. Cognitive psycholo- magnitude shorter than his argument would gists and cognitive neuroscientists have repeat- require (e.g., Sperling, 1960). His interpreta- edly made discoveries that either violated tion also has no empirical support: no evidence prevailing intuitions or that intuition never for prolonged sensory persistence has been would have contemplated: recent examples provided by any studies of sensory processes include the evidence for implicit memory. for in infants. These considerations have no force. multiple representations of objects, and for however, if Haith’s default rule is accepted. separate visual coding of surface color and Even the most implausible and unsupported motion. If human intuitions are not a trustwor- sensory interpretation of data from infant stud- thy source of knowledge about the cognitive ies is preferable, by this rule. to any cognitive processes of adults, they are hardly likely to be interpretation. more trustworthy guides to knowledge about The requirement that claims of cognitive cognition in infants. The intuitions and precon- competence be proved by the elimination of ceptions of scientists can never be eliminated every alternative claim, however implausible from science, but they should not be used to and unsupported, sets an impossible standard filter the evidence that research brings. for research on cognition in infancy. Like any Related to this skeptical reaction is a ten- other branch of science. the study of cognitive dency to judge the findings of studies of cogni- development is not an exercise in logic reault- tion in infancy against an impossible standard. ing in irrefutable conclusions: Hypotheses can For example. Haith (1997) claims that investi- be rejected or supported by evidence but can gators who use preferential looking methods to never be proven correct. Because there are an probe infants’ cognitive capacities “must fend infinite number of alternative interpretations of Nativism, Empiricism, and Know/edge 191 any finding in any area of science, empirical helpful to the field when the accounts are progress requires that scientists select and developed in a principled manner and tested by evaluate interpretations in accord with evi- further research. Healthy progress has come, dence, not in accord with a priori preferences for example, from Cohen’s studies of the for some interpretations over others. No sources of infants’ reactions to violations (and hypothesis can be held to be true or false until interesting non-violations) of object solidity proven otherwise. (Cohen, 199.5; Cohen, Gilbert, & Brown, A third problem facing investigators of cog- 1996) and Oakes’ studies of limits to infants’ nition in infancy is a tendency of some critics sensitivity to contact-mechanical motions to consider individual studies in isolation, (Oakes, 1994) and to gravity (Kannass & rather than to develop unitary and principled Oakes, 1997). Further progress may come accounts for a larger body of research. One from Bogartz’s new methodological and statis- example of this tendency is discussed in foot- tical approaches to preferential looking note 3. As a second example, Haith’s (1997) research, although the sensitivity of these suggestion that sensory persistence accounts approaches remains to be demonstrated (see for apparent cases of object representation is footnote 3). Finally, progress is coming from framed in the context of a discussion of studies studies revealing surprising limits to infants’ in which infants view stationary objects that representations of occlusion events (e.g., lirst are fully visible and then are occluded for Chiang & Wynn, 1996; Huntley-Fenner & several seconds (e.g., Baillargeon & Devos, Carey, 1995; Xu & Carey, 1996). To advance 1991; Wynn, 1992). This suggestion cannot understanding of early cognitive development, account for the findings of numerous studies those who are skeptical of current accounts of presenting much longer occlusion times (e.g., cognition in infancy should not ignore their Baillargeon & Graber, 1988; Wilcox, Rosser, skepticism but submit it to test, adhering to & Nadel, 1994) occluded objects that move or guidelines that all investigators can follow. I change (e.g., Rochat & Hespos, 1996; Koech- suggest four guidelines: lin, Dehaene & Mehler, 1997; Simon, Hespos & Rochat, 1995), or objects with surfaces that Theories should be evaluated in rela- are never visible (e.g., Johnson & Aslin, 1995; tion to evidence, not compatibility with Kellman & Spelke, 1983; Van de Walle & intuition. Spelke, 1996). Although separate explanations No hypothesis should be considered could be proposed for the findings of each of “guilty until proven innocent” or the these studies, our understanding of infant cog- reverse. nition is not likely to advance if we propose All accounts of the findings of infant new explanations for each new set of findings. studies require evidence. In particular, Requiring all rival accounts of cognitive devel- those who would explain infants’ per- opment to be responsive to all experimental formance by appealing to sensory or findings would help to place discussions on a motor processes must provide evidence firmer foundation, focusing attention on areas for those processes, on a par with those where serious alternative explanations exist who would explain infants’ perfor- and where further research would be most pro- mance in terms of perceptual or cogni- ductive. tive processes. I do not claim that every study of perception All theories of early cognitive develop- and cognition in infancy has been correctly ment must encompass all the relevant interpreted by its authors, or that every skepti- data. In particular, explanations of cal reaction to this research impedes progress. infants’ performance that appeal to sen- On the contrary, the development of compet- sory-motor processes, motivational ing accounts of findings can be extremely processes, perceptual processes, and 192 INFANT BEHAVIOR X DEVELOPMENT Vol. 21, No. 2, 1998
cognitive processes must all be held to independently of perceptual encounters with the same standard; no account merits those scenes and objects? These questions are attention if it is based on a small subset not addressed by research on interactions of findings and ignores contrary results. between genes and gene products but by research on the emerging and changing capac- ities of children in interaction with their sur- ARGUMENTS AGAINST NAT/V/ST roundings. CLAIMS Construed appropriately, the questions about the sources of human knowledge are not It is worth asking why the intuitions of many incoherent but well-formed, and some of them investigators have favored extreme empiricist are straightforwardly testable. Psychologists theories and skewed standards for evaluating who study animals can and have asked research. A number of arguments in support of whether a dark-reared rat perceives depth on these intuitions and standards have been first encountering the light, and whether a offered. Here, I consider six arguments against newborn chick represents an occluded object any nativist interpretations of research on cog- the first time it sees an object being hidden. nition in infancy, according to which such Psychologists who study humans can and have interpretations are incoherent, false, unparsi- asked whether a newborn infant with no visual monious, empty, denying of flexibility, or experience perceives depth, distinguishes socially dangerous. In each case, I suggest the faces from other kinds of patterns, or repre- arguments are misplaced, and that the consid- sents occluded objects. Investigators also have erations motivating them should lead investi- asked about the role of specific experiences gators in a different direction. such as locomotion in the development of per- ception and representation: a very fruitful con- tribution to the dialogue (e.g., Bertenthal & Na tivism is Incoherent Campos, 1990). The fascinating advances in As developmental biologists have shown in research in neurobiology do not undermine exquisite detail, all development involves a these questions. At its best, research in neuro- process of interaction between genes and envi- biology suggests mechanisms by which cogni- ronment. Without the right physical and chem- tive structures can develop in advance of ical environment, genes are inert and no sensory contact with the external world, as development happens. From this finding, some well as mechanisms by which these structures developmental psychologists have concluded can be shaped and modified by such contact.’ that it is incoherent to imagine that any knowl- edge of the world could have its source solely Nativism is False in the organism (e.g.. Oyama, 1985; Thelen & Smith, 1994). When the findings of studies of early cogni- The problem with this argument is that the tive development are scrutinized with appro- nativist-empiricist dialogue is not about the priate rigor, some investigators argue, they interaction of genes and their environment, but yield no evidence for knowledge of things pre- about whether knowledge of things in the ceding experience with those things. Rather. external world develops on basis of encounters the evidence suggests that all knowledge with those things. Do we learn to perceive results wholly from dynamic interactions with depth by looking at three-dimensional scenes? the external environment (Elman et al, 1996; Do we learn to see objects by looking at and Munakata et al. 1997: Thelen & Smith, 1994). manipulating objects‘? Alternatively, do struc- This conclusion rests in part on skewed tures for representing three-dimensional interpretations of studies of cognition in scenes furnished with bounded objects develop infancy, as discussed above. and it is further 1Y.i nourished by a general error of interpretation should not be discouraged from conducting of developmental data. Faced with evidence research to reduce that ignorance by skewing for a developmental change in some capacity, their interpretations of the evidence already at investigators are apt to conclude that the cause hand. of the change is learning, ignoring two alterna- tive possibilities. First, the capacity may be constant over development but the ability to Nativism is Unparsimonious express it may change because of other devel- Some investigators have granted that ques- opmental changes (see Banks & Shannon, tions about the origins of knowledge are mean- 1988; Thelen, 1984, for examples). Second, ingful and empirical. Because existing the capacity may emerge over development research does not yet resolve these questions in but the cause of its emergence may be matura- many cases, they argue, the most parsimonious tion or triggering rather than shaping by expe- assumption is that knowledge is lacking early rience (see Held, 1985, for an example). This in development. Until the evidence forces one error of interpretation fosters the conclusion to a different conclusion, on this view, one that knowledge has been acquired through should assume that young infants lack all learning when all that is known is that behavior knowledge and cognitive processes. on some task has changed. This argument rests, I believe, on misunder- Instead of drawing empiricist conclusions standings of the role of parsimony consider- automatically, students of cognitive develop- ations in science and of the nature of ment should conclude that learning has taken developmental theories. First, parsimony is place only when there is evidence for learning, appropriately invoked in cases where a rich from research revealing that different knowl- body of evidence is consistent with two or edge emerges under different environmental conditions. If one bases conclusions only on more detailed theories. When evidence is evidence, then I believe that studies of infants sparse and theories are sketchy, as in the study suggest that development is not strongly of cognitive development, scientists need to skewed toward either pole of the nativ- collect further evidence. not jump to conclu- ist-empiricist dialogue. There is some evi- sions on grounds of parsimony. Second, theo- dence for innate knowledge, embodied in ries of development aim to describe and structures that develop in advance of their explain how the capacities of adults come to function and in advance of relevant perceptual be. Parsimony arguments apply to these theo- contacts with the objects of knowledge. (This ries as n~holes: The most parsimonious theory evidence seems to me strongest in the cases of of cognitive development is the theory provid- depth perception and face processing.) There ing the simplest account of the development of is also some evidence for learning, from situa- mature knowledge, not the simplest descrip- tions where children’s knowledge varies with, tion of the young infant. Because all theories and because of, variations in their experience. must arrive at the same end state of mature (This evidence seems to me strongest in the knowledge, accounts with simpler character- cases of speech perception and certain spatial izations of the initial state will tend to have representations.) Finally, there are vast areas more complex characterizations of develop- of ignorance, where the contributions of innate mental change. If one focuses on the simphcity structures and learning have not been disentan- of developmental theories as wholes, rather gled. Students of development should not be than the simplicity of the pieces of those theo- surprised or discouraged by the extent of our ries characterizing the initial state, then parsi- ignorance, because the experimental study of mony considerations do not automatically cognition in infancy is a young enterprise and favor one voice over others in the nativ- it is progressing. Above all, investigators ist-empiricist dialogue. lY4 INFANT I3tHAVIOII X DEVELOPMENT Vol. 21, No. 2, lYY8
Nativism is Empty set of “constraints on learning” (e.g., Gelman, 1990; Keil, 198 1; Spelke, 1990). This termi- Perhaps the most common argument nology is in some ways unfortunate, for it against nativist claims is that they do not appears to imply that innate knowledge pre- explain development: To say that a given vents people from learning (see Quartz & aspect of knowledge is innate is not to account Sejnowski, 1997). In fact, innate structures for its emergence or its form. Nativist claims, it have traditionally been proposed in order to is argued, only shift the burden of explaining explain how it is possible for humans to learn development to some other discipline, such as anything. They do not deny human flexibility developmental biology. but instead participate in attempts to under- This argument misconstrues the nature of stand both human flexibility and its limits. explanation in developmental psychology. All For example, theories positing initial theories of cognitive development have the knowledge have been proposed to explain how dual task of characterizing the initial state of is it possible for human children to learn any knowledge and the processes that transform human culture’s language, motor skills, and this initial state into mature knowledge. In object taxonomies, or formal belief systems extreme empiricist theories, the initial state (e.g., Chomsky, 1975; Hirschfeld & Gelman. typically is held to consist of a set of innate 1994). Theories that posit unlearned systems sensory transducers and one or more mecha- of knowledge have even been proposed to nisms of learning; in other theories, the initial account for the development of humans’ most state and developmental mechanisms are char- flexible, formal belief systems (e.g., Carey & acterized differently. Because all theories Spelke, 1994, 1996: Sperber, 1994). Debates across the nativist-empiricist spectrum have between nativists and empiricists are not deni- the same general form, the explanatory value als and assertions of flexibility but contrasting of each theory depends only on how well it accounts of the sources and the nature of accounts for the phenomena of development humans’ often flexible cognitive performance. and on theory-internal qualities such as com- pleteness and consistency. A theory’s explana- Na tivism is Dangerous tory value does not depend on the content it assigns to the initial state. Perhaps the most serious argument against To build good explanatory theories, stu- nativist claims focuses on the impact of these dents of cognitive development must seek the claims on society. The thesis that certain sys- most complete, consistent, and empirically tems of knowledge are innate in our species is adequate account of the initial state and subse- sometimes said to go naturally with the thesis quent growth of knowledge. As psychologists that some people are inherently more capable learn more about cognition in infancy, the con- thinkers and knowers than others. As is well straints on all theories grow and the explana- known, this second thesis has underpinned tory virtues of different theories will become social evils such as racist immigration policies, clearer. Developing better explanatory theories it serves to rationalize social injustice. and it requires vigorous programs of research threatens to foster further, regressive social addressing the questions at the center of the changes. By this argument, nativist claims nativist-empiricist dialogue: it is not aided by (I should be shunned so as to avoid these social priori rejection of one side of the dialogue. consequences (Elman et al. 1996; Fischer & Bidell. 1994). Na tivism Denies Flexibility The problem with this argument lies in its first premise: The question whether any Investigators of cognitive development knowledge is innate in our species is entirely sometimes characterize initial knowledge as a different from the question whether there are Ndlivism, Empiricism, and Knowledge any innate differences between people in ple are salient. Studies of the origins and early knowledge or cognitive ability. Consider, for development of knowledge serve to increase example, a scientist who believes that a system awareness of the vast common ground uniting of knowledge of objects is innate in all people, all human thinkers, helping us to understand and who asks why adults differ in the extent to what it is to be a human thinker and knower in which they go beyond this system: why one any culture and in any set of circumstances. student of physics gets an A whereas another Much of the heat in the controversies over IQ gets a C, or why one athlete-in-training consis- and multiculturalism may dissipate as this tently hits baseballs whereas another consis- understanding grows. tently misses. It is completely open to this Research guided by the nativist-empiricist scientist to believe that all differences between dialogue does not, however, deny human dif- people stem from differences in their experi- ferences. On the contrary, it sheds light on the ences: their differing opportunities to extend particular circumstances that lead different their knowledge and abilities in classrooms or people to extend their knowledge and skills in on athletic fields. Consider now a second sci- different directions. Where knowledge is entist who believes that all knowledge of found to vary across people in different cul- objects is learned and who asks the same ques- tures or circumstances, that variability teaches tion about the sources of individual differences us something about our own potential and that in adults. It is entirely open to this scientist to of others. This information can guide choices believe that differences among physics stu- about how to educate children and structure dents and baseball players stem from differ- societies, and it can help everyone to view the ences in people’s innately given learning differing accomplishments of different people capacities.5 For better or worse, claims about with understanding and respect. the sources of the knowledge that all people These are not new reasons for asking about share do not bear questions about the sources the origins and growth of knowledge, for they of the abilities that distinguish one person from trace back to the beginnings of the nativ- another. ist-empiricist dialogue. What is new are the advances in cognitive science that now allow students of cognitive development to address THE NAT/V/ST-EMPIRICIST DIALOGUE these questions empirically. By pursuing that IN A LARGER SOCIAL PERSPECTIVE work and overcoming old prejudices, our understanding of human knowledge and Although studies of cognition in infancy do human nature may grow considerably in the not reveal the sources of individual differences coming years, enriching and informing in ability or achievement, I believe that they long-standing social dialogues on human cast a valuable new perspective on those dif- nature, human differences, and human devel- ferences. When experiments reveal systems of opment. knowledge that emerge early in human devel- opment and that persist and grow in common Acknowledgment: This article is adapted ways over all children, they suggest that the from portions of a chapter written with Elissa cognitive differences between people are not L. Newport (Spelke & Newport, 1998), and as great as many current discussions imply. from an invited debate at the meeting of the Debates over the genetics of IQ and over cul- Society for Research in Child Development, tural differences in language and thinking tend Washington, D.C., April, 1997. I thank the to overlook the cognitive capacities and attain- many participants of the SRCD meeting who ments that all people share, because most of commented on the debate, and Yuko our common cognitive endowment is obscure Munakata and Fei Xu for comments on an to intuition whereas differences between peo- early version of the manuscript. 1 ‘HI INFANT I3tHAVIOK cy,DEVELC WMtNT Vol. L 1, No. 2, lcJ’lil
NOTES observe novelty prel’errnces for one fully visi- ble object alter habituation to another fully cis-
I. Although Piaget’x (1953) theory of the devel- ibte object. however. the absence of novelty opment of object representation cannot be precercnces after habituation to I cen- placed clearly within the framework of’ the ter-occluded object cannot be taken either- as evidence for any specific limitation to inflints‘ nativist-empiricist dialogue. Piaget also perception of partly occluded objects or as any emphasized the importance of’ object-directed challenge to the findings of Kellman (Kellman actions for the development of ob.ject know I- & Spelke. 1983. Keltman et al. 19X6, IYX7). edge. Johnson (Johnson Kc Aslin, l9Y5, lYY6; 2. Kellman and Arterberry (in press) and Spelkc Johni;on B Naliez, 1995). Jusc/yk et al (I 997,. and Newport ( 19%) offer more complete Needham ( I YY4). or Slatcr et al ( I YYOb). reviews of this research. Kcllman and Artcr- 4. Spelke & Newport (IYYX) discuss possible berry (in press) and Spelke and Newport ncurobiological mechanisms For the dc\elop- (19YX) offer more complete ret iews of this ment of object representation\. research. 5. Claims that all knowledge ia learned frccluentlq 3. Rogart/ and Shinskcy (in press) recently accompany claims that difl’crences in cogniti\c reported a divergent finding. Like the five sets ability are innate; see Herrn\tein and Murra). of’ investigators cited above. f3ogart/. and Shin- ( I YY3) for a recent exuniplc. skey habituated one group of infants to a ccn- ter-occluded object and then tested them with a fully visible continuous object and with 3n REFERENCES dject with a gap. In contrast to the inl’ant~ in the above studies. these infants showed equal looking times to the two test displays. In f’ur- Ahmecl. A. & Kull’tnan. T. ( lYY6). Do inf’ants know ther conditions similar to two control condi- when they are searchin g incol-redly’! Looking tions reported by Kellman & Spelkr ( 19X3. timea in ;I non-hearch A not 13 task (abstract). Experiments 2 and 1). f3ogartL and Shinxkcy //~/il/l/ Brlltrl~ior~ ~1111/t~c'\~c'lol,,llcrlt. IO, 297. habituated tw’o f’urther groups of’ inl’ants either Arterberry. M. E. ( t YY.3). Devclopmcnt 01‘ \pa- to a fully visible continuou‘; djcct or to ;I I’ully tintemporal integralion in inflincy, /rr/tr/i/ Hc,lfrr~,- visible object with ;I gap and then tested them ior. ~IIld I~c,~~rlo/"'rc,,lt. Ifi. 343-363. with the ame two fully visible displayh. In Baillqcon. R. ( 19X6). Representing the exi\tcncc contrast to the inl’ants in Kcllman KL Sprlke.5 and the location 01‘ hidden ob,ject\: Object pel-- ( 19X.1) control conditions. lhese inl‘;int\ aI50 manencr in 6 and X-month-old inI;mt\. Co,q,~i- showed equal looking times to the two test cfic- tio,1. 2.3. 2 I-4 I plays. Bogart/ and Shinsky bad thcit- discus- I3aillargeon. R. ( IOY)?). The object concept rc‘\is-
sion only on their own findings and those 01 iled: Nru direction\ in the invcztigation 01 one condition of Kelltnan Kc Spelke’s ( 10X3) inl’ant\’ physical knouledge. In C. E. Granrud
I’irst cxpcriment. without citing any othrl (Ed.), Vkrrtrl pcrc~~piio~~trmi c~o,qttiliottin irtfirrn. experimental conditions or investigators. They (pp. 265.1 16). Hillsdale. NJ: Erlbaum. suggested that inf’unts fail to show novelty prcl‘- Baillargeon, Ii. & DeVoa. J. ( IYY I ). Object perma-
erencrs when tcstcd with the method and dih- nence in young int’ants: Further evidence. Child plays of‘ Kellman and Spelkc. but the I’indings f)c,r,c,/o//,/rrc,rtr.62. 1227. 1236. reported by Kellman 6i Spelke (lY83. Expel-i- 13aillargeon. R. & Grabcr. M. (IYXH). Evidence 01‘ ment\ 2, 3. and 3). Slater et al ( IYYOb,f‘:xpet-i- location mcmol-y in X-month-old infants. c‘o,q,ti- ment 4). Johnson & A\lin ( lY96. Experiments riort.20. I Y I-208. I and 2). and Nerdham ( I YY4) provide e\ i- Ball. W. A. ( lY73. April). 711~ /wr'c,cptio/tc~/'curr.vtrl- dence against this suggestion. It iy not clca if>,irr //ic ir!/itrzr. Paper presented at lhe merling why Bogart/. and Shinshey’ method failed to 01‘ the Society l’or Research in Child Develop- elicit novelty preferences: their use 01’ small ment. Philadelphia. PA. numbers of inflInt\ and test trial<;, a larger nun- I3anha. M. S.. Kc Shannon. I,. (lYY3). Spatial ;111d
ber of’ test stimuli, XKI older inl’ants are tlil’l’er- chromatic visual cl’ficicncy in human neonaic\. cnca Lvorthy 01‘ test. fJecnu\c they did not In C. I<. Granrtid (Ed.). l'i.vrfolpct~~rpliort ~rtit/ Nativism, Empiricism, and Knowledge 197
cognition in infancy (pp. l-46). Hillsdale, NJ: dam, North Holland: Elsevier Science Publishers Erlbaum. B. V. Berkeley, G. (1709/1975). An essay toward a new Dehaene-Lambertz, G., & Dehaene, S. (1994). theory of vision. In Philosophical works (M. R. Speed and cerebral correlates of syllable dis- Ayers, Ed.). London: Dent. crimination in infants. Nature, 370, 292-295. Bertenthal, B. I., & Campos, J. J. (1990). A systems Diamond, A. (1990a). The development and neural approach to the organizing effects of self-pro- bases of memory functions as indexed by the AD duced locomotion during infancy. In C. and delayed response tasks in human infants and Rovee-Collier and L. P. Lipsitt (Eds.), Advances infant monkeys. In A. Diamond (Ed.), The devel- in infancy research, (Vol. 6, pp. I-60). Nor- opment and neural bases of higher cognitive wood, NJ: Ablex. Functions. Annuls of the New York Academy of Bogartz, R. S., & Shinskey, J. L. (in press). On per- Sciences, 608, 5 17-536. ception of a partially occluded object in Diamond, A. (I 990b). Developmental time course 6-month-olds. Cognitive Development. in human infants and infant monkeys, and the Carey, S., & Spelke, E. S. (1994). Domain-specific neural bases of, inhibitory control in reaching. In knowledge and conceptual change. In L. Hir- A. Diamond (Ed.), The development and neural schfeld & S. Gelman (Eds.), Mapping the mind: bases ofhigher cognitive Functions. Anna1.s oj Domain-.spec$city in cognition and culture (pp. the New York Academy of Sciences, 608, 169-200). Cambridge: Cambridge University 637-676. Press. Elman. J., Bates, E., Johnson, M., Karmiloff-Smith, Carey, S., & Spelke, E. S. (1996). Science and core A.. Parisi, D., & Plunkett, K. (1996). Rethinking knowledge. Philosophy of Science, 63, 5 1S-533. innateness: A connectionist perspective on Casey, B. J., Trainor, R. J., Orendi, J. L.. Schubert, development. Cambridge, MA: MIT Press. A. B., Nystrom, L. E., Giedd, J. N., Castellanos, Fantz, R. L. (1961). The origin of form perception. F. X., Haxby, J. V., Nell, D. C., Cohen, J. D.. Scientific American, 204, 66-72. Forman, S. D., Dahl, R. E., & Rapoport, J. L. Fischer, K. W., & Bidell, T. (1994). Constraining (I 997). A developmental functional MRI study nativist inferences about cognitive capacities. In of prefrontal activation during performance of a S. Carey & R. Gelman (Eds.), The epigenesis of go-no-go task. Journal of Cognitilre Neuro- mind: Essays on biology and cognition (pp. I99- science. 9, 835-847. 236). Hillsdale, NJ: Erlbaum. Chiang. W.-C. & Wynn, K. (1996). Eight-month-old Friedman, S. (1972). Habituation and recovery of infants’ reasoning about collections (abstract). visual response in the alert human newborn. Infunt Behavior and Development, 19, 390. Journal of Experimental Child Psychology, 13, Chomsky. N. (1975). R
Herrnstein, R. J., & Murray, C. A. (1994). The hell Kellman, P. J.. & Arterberry, M. (in press). The c’ru- ~uT\~P: intelligence und c1a.s.sstructure in Ameri- dle of knowledge: Development of perception in cun l@. New York: Free Press. infunc_y. Cambridge, MA: MIT Press. Hirschfeld, L. A., & Gelman, S. A. (1994). Mupping Kellman. P. J.. Gleitman. H.. & Spelke, E. S. the mind: Domain-specificity in cognition and (19X7). Object and observer motion in the per- culture. Cambridge: Cambridge University ception of objects by infants. Journul of Experi- Press. mentul Psychology: Human Perception und Hofsten. C. von, & Spelke, E. S. (1985). Object per- Perjbrmuncr. 13, 5X6-593. ception and object-directed reaching in infancy. Kellman. P. J., & Spelke, E. S. (1983). Perception of Journul of Experimentul Psychology: Genrrul, partly occluded objects in infancy. Cognitil~r 114, 39X-212. P.s~cholog~. 1.5. 4X3-524. Hofsten, C. von, Spelke. E., Feng, Q.. & Vishton. P. Kellman, P. J., Spelke, E. S., & Short, K. R. (1986). (1994). Infants’ predictive head turning and Infant perception of object unity from transla- reaching for fully visible and occluded objects tory motion in depth and vertical translation. Child De\~elopmc~nt. 57, 12-X6. (abstract). Infunt Behaviorund De~~elopmc~nt.17. 1000. Koechlin, E.. Dehaene, S. & Mehler, J. (in press). Numerical transformations in five-month-old Hood, B., & Willats, P. (IYX6). Reaching in the dark human infants. Muthcmatical Cognition. to an object’s remembered position: Evidence for object permanence in S-month-old infants. Lea, S. E. G., Slater, A. M., 6i Ryan. C. M. 6. (1996). Perception of object unity in chicks: A British Journul of Dn~elopmmtal Psychology, 4. 57-65. comparison with the human infant. Irtfunt Brhu~~ior and Dc~~~rlopmc~nt,19, SO I-504. Huntley-Fenner, & Carey, S. (1995, April). Physi- Leslie, A. M. (I 994). ToMM, ToBy, and agency: cul rcwsoning in infuncy: The wprr~srntution of Core architecture and domain specificity. In Hir- nonsolid substuncrs. Paper presented at the schfeld, L. A. & Gelman, S. A. (Eds.), Mapping meeting of the Society for Research in Child thr mind: I)omain-sp~cificit!: in cognition and Development, Indianapolis, IN. c.ulturc (pp. I 19-148). Cambridge: Cambridge Johnson, S. P., & Aslin, R. N. (1995). Perception 01 University Press. object unity in 2-month-old infants. Dpvclop- Leslie. A. M. & Keeble, S. (1987). Do six-month- rncntul P.s~cholo~~. 31. 739-745. old infants perceive causality‘? Cognition. 25, Johnson. S. P., & Aslin, R. N. ( I YY6). Perception of 265-288. ob.ject unity in young infants: The roles of Mandler. J. M., & McDonough. L. (19Y3). Concept motion. depth. and orientation. Cogniti\ae De\‘cI- formation in infancy. Co,qnitivr De~~elopmrnt. 8. opmrnt. I I, I6 I- I HO. 291-31X. Johnson, S. P.. & Nafiez, J. E. Sr. (IYYS). Young Munakata, Y. ( I YY7. April ). il .sirrgle. graded system infants’ perception object unity in two-dimen- of c!f kno~~lrdgr: The pocc’rr of‘~craknes.s to expluin hional displays. Irziirnt Bchu\~ior und Dr~~elop- tLI.sk-dL’/)ClldPnthehulaior. Paper presented at the 177<‘/11. IK. 13% 13.3. meeting of the Society for Research in Child Juhczyk, P., Johnson. S., Spelke, E.. Kennedy, L., & Development, Washington, D.C. Smith, C. lY97, April). ( Does .synchronon.s Munakata, Y.. McClelland, J. L.. Johnson, M. H.. Kc (hflll,@ o,,e,’ time .spcTcif_i 0bjec.t unifi ,fijr Siegler, R. S. (lYY7). Rethinking infant knowl- ir!firnts? Poster presented at the meeting of the edge: Toward an adaptive process account 01‘ Society for Research in Child Development, successes and failures in object permanence Washington, DC. tasks. P.s~c~llolo~~ic.~~IRe\sirbv. 104. 6X6-7 13. Karmas\. K. N.. & Oakes, L. M. (IYY7, April). Needham, A. ( lYY4). Infants’ use of perceptualsim- Downhill \‘.s.uphill: Irzfirnts smsitil~ity to grmi- ilarity when segregating partly occluded objects tutiom~l cwwtruints 011object motion. Poster pre- during the li)urth month of lilt (abstract). lnfilnt sented at the meeting of the Society for Research Behavior and Dcvrlopmc~nt. 17. 163. in Child Development, Washington, D.C. Needham, A. (lYY7). Factors affecting infants’ use Keil, F. C. (IYX I ). Constraints on knowledge and of featural information in object segregation. cognitive development. Psycholo~icrrl Rc\sieM., Current Dircjc,tion.s in P.syclzolo~icrrl .Scicnc(,. 6. (IK. I Y7-227. 26-33. Na livism, Empiricism, and Know/edge 199
Needham, A., Baillargeon, R., & Kaufman, L. Spelke, E. S. (1990). Principles of object percep- (1997). Object segregation in infancy. In C. tion. Cognitive Science, 14,29-56. Rovee-Collier & L. Lipsitt (Eds.), Advances in Spelke, E. S., Breinlinger, K., Jacobson, K., & Phil- infancy research (Vol. 11, pp. l-44): Norwood, lips, A. (1993). Gestalt relations and object per- NJ: Ablex. ception: A developmental study. Perception, 22, Noland, J. S. (1996). Inhibitory control and visually 1483-1501. guided detour reaches (abstract). Infant Behavior Spelke, E.S., Hofsten, C. von, & Kestenbaum, R. and Development, 19,65 I. (1989). Object perception and object-directed Oakes, L. M. (1994). The development of infants’ reaching in infancy: Interaction of spatial and use of continuity cues in their perception of cau- kinetic information for object boundaries. Devel- sality. Developmental Psychology, 30, 869-879. opmental Psychology, 25, 185-196. Oyama, S. (1985). The ontogeny of information: Spelke, E.S., Katz, G., Purcell, S. E., Ehrlich, S. M., Developmental systems and evolution. Cam- & Breinlinger, K. (1994). Early knowledge of bridge: Cambridge University Press. object motion: Continuity and inertia. Cognition Piaget, J. (1952). The origins of intelligence in 1,131-176. childhood. New York: International Universities Spelke, E. S. & Newport, E. L. (1998). Nativism, Press. empiricism, and the development of knowledge. In W. Damon (Series Ed.) & R. Lerner (Vol. Piaget, J. (1954). The construction of reality in the Ed.), Handbook of Child Psychology: Vol. 1. child. New York: Basic Books. Theoretical Models of Human Development. Quartz, S. R. & Sejnowski, T. J. (1997). The neural (5th ed., pp. 275-340). New York: Wiley. basis of cognitive development: A constructivist Spelke, E. S. & Van de Walle, G. A. (1993). Per- manifesto. Behavioral and Brain Sciences, 20, ceiving and reasoning about objects: Insights 537-596. from infants. In N. Eilan, W. Brewer, and R. Regolin, L., & Vallortigara, G. (1995). Perception McCarthy (Eds.), Spatial representation (pp. of partly occluded objects by young chicks. Per- 132- 16 1). Oxford: Basil Blackwell. ception and Psychophysics, 57.97 l-976. Sperber, D. (1994). The modularity of thought and Regolin, L. Vallortigara, G., & Zanforlin, M. the epidemiology of representations. In L. Hir- (1995). Object and spatial representations in schfeld & S. Gelman (Eds.), Mapping the mind: detour problems by chicks. Animal Behavior, 49, Domain-specificity in cognition and culture (pp. 195-199. 39-67). Cambridge: Cambridge University Rochat, P. & Hespos, S. J. (1996). Tracking and Press. anticipation of invisible spatial transformations Sperling, G. (1960). The information available in by 4- to 8-month-old infants. Cognitive Drvelop- brief visual presentations. Psychological Mono- ment, II, 3-17. graphs, 74 ( 11). Simon, T., Hespos, S., & Rochat, P. (1995). Do Termine, N., Hrynick, T., Kestenbaum, R., Gleit- infants understand simple arithmetic? A replica- man, H., & Spelke, E. S. (1987). Perceptual tion of Wynn (1992). Cognitive Development, completion of surfaces in infancy. Journal of IO, 253-269. Experimental Psychology: Human Perception Slater, A., Johnson, S. P., Kellman, P. J., & Spelke, and Perjcjrmance, 13, 524-532. E. S. (1994). The role of three-dimensional Thelen, E. (1984). Learning to walk: Ecological depth cues in infants’ perception of partly demands and hylogenetic constraints. In L. P. occluded objects. Early Development and Lipsitt & C. Rovee-Collier (Eds.), Advances in Parenting, 3, 187- 19 1. Infancy Research (Vol. 3, pp. 214-250). NY: Slater, A., Mattock, A., & Brown, E. (1990). Size Ablex. constancy at birth: Newborn infants’ responses Thelen, E. (1995, April). A dymamic systems to retinal and real size. Journal of Experimental approach to the object concept. In L. B. Smith Child Psychology, 49, 3 14-322. (Chair), Tests of a dynamic systems theory: The Slater, A., Morison, V., Somers, M., Mattock, A., object concept. Symposium conducted at the Brown, E., & Taylor, D. (1990). Newborn and meeting of the Society for Research in Child older infants’ perception of partly occluded Development, Indianapolis, IN. objects. Infant Behavior and Development, 13, Thelen, E. & Smith L. B. (1994). A dynamical sys- 33-49. tems approach to the development of cognition 200 INFANT BEHAVIOR & DEVELOPMENT Vol. 21, No. 2, lYY8
and u&on. Cambridge, MA: Bradford/MIT at the meeting of the Society for Research in Press. Child Development, Washington, DC. van Giffen, K., & Haith, M. M. (1984). Infant visual Wynn, K. (1992). Addition and subtraction in response to gestalt geometric forms. /nj&7t infants. Nature, 358, 749-750. Behavior and Dwelnpment, 7. 335-346. Wynn, K. (1995). Infants possess a system ot Van de Walle. G. A. & Spelke E. S. (1996). Spa- numerical knowledge. Current Directions in tiotemporal integration and object perception in PxyholoRical Science, 4, 172-117. infancy: Perceiving unity vs. form. Clrild Drwl- Xu, F., & Carey, S. (1996). infants’ metaphysics: opment. 67, 262 I-2640. The case of numerical identity. Cqtzitive Psy- Wilcox, T., Rosser, R., & Nadel. I>. (I 994). Repre- cholo~~, 30, I I I- 153. sentation of object location in 6.5-month-old Yonas, A., & Granrud, C. ( 1985). Development 01 infants. Cognitive Development, 9, 193-2 IO. visual space perception in young infants. In Woodward, A. (1995, April). Irzfants rea.wning J. Mehler and R. Fox (Eds.), Neonate u~gnition: trhocct the goals of a human actor. Paper pre- Brymd the blooming. buzzing confusion (pp. sented at the meeting of Society for Research in I I3- 145). Hillsdale. NJ: Erlbaum. Child Development, Indianapolis. IN. Wu, X. (I 991, April). Devrlopmrntal c~hungc’.sin ir+u~ts’ individuation of‘pwplr. Paper presented