Symposium
Chairs: Nora S. Newcombe and Kathryn Hirsh-Pasek
Discussant: J. Steven Reznick
The aim of cognitive developmental research is to understand theinteraction of biological and environmental influences in theformation of mature capabilities. Although almost all theoristsagree with the general (but vacuous) claim that there isinteractionism in development, different theorists give differentemphasis to biological and environmental factors, and proposevery different accounts of development. The purpose of this symposium is to present an overview of recentinvestigation and theorizing regarding three central aspects ofcognitive competence in infancy: language, space and quantity. Ineach of these domains, there have been proposals that infantscome equipped with content-specific knowledge structures and/ormodules, with the role of the environment and of domain generalprocessing being regarded as relatively unimportant. The work tobe presented in this symposium differs from such proposals inthat it attempts to encompass both evidence of strong startingpoints for development in the endowment of the infant, and alsoevidence that experience in the physical and social world isvital to development. Each presenter does so by proposingparticular hypotheses about the nature of interactions of theinfant with the world. Such efforts are vital to ensure thatinteractionist approaches are empirically testable. The presenters and discussant will also consider: (a) thecommonalities and differences among these three domains, and (b)what we learn from these research efforts about how better tocharacterize what is meant by interaction and change.
Details of individual items:
paper
Psycholinguistics has a long history of adopting myopic solutionsto the complex problem of language acquisition. Though many paylip service to interactive theories of development, most explainlanguage growth through appeal to nativistic, or social, orcognitive sources of information. Here, we propose an integrativeapproach that incorporates each of the available theories,offering evidence that children rely on the interaction ofmultiple cues to solve the logical problems of languageacquisition (Hirsh-Pasek & Golinkoff, 1996; Hollich, Hirsh-Pasek& Golinkoff, in press; Hollich, 1999). Children are predisposedto note correlations of certain inputs over others across anumber of domains. For example, infants are attuned to prosodiccues like vowel lengthening, to social inputs like eye gaze, andto perceptual cues like novelty. As they differentiallyintegrate multiple sources of information over time, theydiscover reliable ways to segment the sound stream, buildcategories of knowledge, and map words and grammar onto objects,actions and events.The case of word learning provides an illustrative model. Anumber of theories address the question of how infants learntheir first words: from social pragmatics (Akhtar & Tomasello, inpress), to constraints (Markman, 1989), to attentional learning(Smith, in press). However, the Emergentist Coalition Modeloffers an empirically testable hybrid account that rests on threeassumptions: 1) infants process multiple sources of information;2) these inputs are differentially weighted over time; and 3)infants move from immature to mature principles of word learning.Both experimental and computational research suggest that thismodel better speaks to the research questions than do any of thetheories considered individually. For example, in onecross-sectional study, 12-, 19-, and 24-month-old infants weretested on their sensitivity to social cues and perceptual cues inmapping words onto objects. Using the interactive intermodalpreferential looking paradigm (IIPLP) (Hollich, Hirsh-Pasek &Golinkoff, 1999), infants saw a boring and an interesting novelobject presented side-by-side on a display board. In onecondition, the boring object was labeled using a social cue, suchas eye gaze. Infants visual fixation revealed that althoughinfants are sensitive to social and perceptual cues at all ages,the youngest relied on perceptual salience while older subjectswere more attuned to subtle social cues in attaching word toobject. This pattern of results held across a wide range ofsocial cues (handling, pointing) and was evident even when thetest objects were of equal salience. Similarly, computational research better specifies the nature ofa truly interactionist account (Hollich, in preparation). Forexample, using a competition-type neural network, Hollich (1999)tested a best approximation of the combination of attentional,constraints, and social accounts. Results suggested that learningis more efficient and reliable when there is some biased learningand when children can use input from multiple sources todetermine word to referent mappings.In sum, both types of data converge to suggest that the changingnature of word learning is a product of infants increasingability to use multiple interacting cues over time.
paper
Several investigators have proposed that spatial developmentbegins with an innate ability to code space in Cartesian terms aswell as to conduct spatial inference (Landau & Spelke, 1981,1984) and, somewhat differently but in the same vein, thatinfants are born with a modularized ability to code geometricshape, with the limitations of this module overcome only with theadvent of relatively sophisticated spatial language (Hermer &Spelke, 1994, 1996). These studies have revealed evidence ofstriking and early-developing spatial skill (see alsoHuttenlocher, Newcombe and Sandberg, 1994). However, suchevidence of strong starting points for spatial development needsto be situated within an approach to spatial development thatalso recognizes the substantial changes infants and toddlers gothrough in their spatial coding and reasoning, not all of whichare due to the acquisition of spatial language. There are manysuch changes, and in this talk, I will discuss several. One important kind of change involves differentiation. While theability to encode distance in continuous space appears to bepresent as early as 5 months (Newcombe, Huttenlocher & Learmonth,1999), this initial skill needs to be differentiated from otherdomains for which such perceptual ability also serves as thestarting point. Specifically, it appears to be the same kind ofperceptual ability that underlies infants initial judgments ofquantity that are later the basis for an idea of number, a matterto be discussed in this symposium by the speaker on quantitativedevelopment. Further, this kind of spatial coding also isinitially undifferentiated from infants ideas of what an objectis (Newcombe et al., 1999; Xu & Carey, 1996). Objecthood for theyoung infant is defined by spatiotemporal trajectory, rather thanobjects having a spatial location as one of their attributes.Differentiation of the spatial aspects of what it means to be anobject from the nonspatial aspects might be driven by variousmechanisms, including induction and/or early labeling experienceswith adults, a matter to be discussed in this symposium by thespeaker on language development.Another important change in the spatial domain involves theadvent of use of codings of distance to fix an object s locationusing distal landmarks. The use of an external framework toestablish location is the most flexible and powerful of humanspatial coding systems. Yet, there is evidence that this kind ofcoding (sometimes termed place learning) is not evident untilabout 21 months of age (Newcombe, Huttenlocher, Drummey & Wiley,1998) and is not at mature levels until 7 years of age (Overmanet al., 1996). In this talk, I will discuss two possible accountsof this developmental course. One emphasizes maturation of thehippocampus, the brain area known to subserve this kind ofspatial coding from animal work, while the other emphasizesexperience with locating (and not locating) objects asenvironmental feedback that progressively refines the earlyskill.Lastly, in this talk I will touch upon the evidence that younginfants initially are disposed to use egocentric or sensorimotorcodings of spatial location. This tendency is sometimesconceptualized as an absolute or exclusive means of spatialcoding in the infant, but I argue that this tendency is betterthought of as a preference for one of several kinds of spatialcoding that co-exist in the infant. There is a considerableamount of empirical work that indicates that locomotory andvisual experience is vital in infants developing a reducedreliance on sensorimotor coding. Such evidence indicates oneimportant role for the environment in spatial development.Although this kind of experience is virtually universal, andhence results in nearly universal sequences of development, itsimportance should not therefore be underestimated.In summary, there is evidence both for strong early startingpoints in spatial development and for considerable developmentalchange along several lines. What remains to be better understoodare the reasons for many of the observed developmental changes.The best worked out case is that of the trend away from relianceon sensorimotor coding, an example that reminds us of the likelyrole of universal experience in the ontogeny of basic aspects ofhuman cognitive competence.
paper
Evidence of quantitative awareness in infants has led some topropose that humans are born with number-specific knowledgestructures (Gallistel & Gelman, 1992; Starkey, 1992; Wynn, 1997). In these accounts, the strengths and limitations of the inbornstructures are emphasized while the roles of domain generalprocessing and environmental input are downplayed. Although theimportance of interactions with the environment are acknowledged,the details of these interactions are rarely discussed. In mytalk, I will seek to identify specific interactions that drivechange in quantitative concepts.First, I will present evidence that the starting point forquantitative development is general perceptual processing ratherthan number-specific structures. Recent research has shown thatinfants use estimates of contour length and surface area, ratherthan discrete number, to respond in quantitative tasks(Clearfield & Mix, 1999; Feigenson & Spelke, 1998). This makessense because scanning edges and surfaces is a fundamental wayinfants can interpret visual scenes. Thus, quantitativeknowledge could emerge from early attempts to analyze the visualworld.Such a starting point implies that infants' first ideas aboutquantity are not based on discrete number. Instead number andoverall amount are united in an undifferentiated sense ofquantity. Eventually, children must discover that number andamount are distinct. In my talk, I will discuss two possiblemechanisms by which this discovery could take place. Onepossibility is that it occurs before children learn conventionalskills simply through experience manipulating sets of objects. Alternatively, children may need to learn and apply conventionalunit measures and counting to make this distinction.Another major accomplishment of early childhood is recognition ofnumerical equivalence. Interestingly, it is many months beforechildren's equivalence judgments encompass a range of diversecomparisons (Mix, in press). I will present evidence related toseveral interactions that might promote this abstraction. First,the earliest numerical comparisons children can make are betweennearly identical sets. This suggests that, as in other domains,numerical abstraction builds from opportunities to compareentities that share irrelevant surface features [i.e., Kotovsky &Gentner's (1996) progressive alignment hypothesis]. Second,children who know the conventional count words can make a widerrange of numerical comparisons. For object sets, it appears thatthe count words serve to highlight the similarity between sets,just as labels do in other comparisons (e.g., Waxman & Markow,1995). For sequential sets, counting is a representationaladvance; it may be the only way children can remember the numberof temporally distributed items. Thus, numerical abstractionresults from the confluence of several environmental andconceptual forces. In summary, although the role of interactions between the childand environment is often acknowledged, the details of theseinteractions are seldom spelled out. The goal of this talk is toidentify some specific interactions in quantitative developmentthat are likely to promote conceptual change.
paper
In this talk I will consider certain themes that cross-cut thecognitive domains covered by the first three speakers : thelanguage, space, and quantitative domains. In all these domains,there are striking similarities in development across children.There are also individual differences among children. However,these differences have not received as much attention frominvestigators of basic cognitive processes as have thesimilarities.I will discuss the possible sources of both the similarities anddifferences among children in early cognitive development. Inboth cases, the sources reflect some combination of thebiological makeup of children and of the environments theyencounter. With respect to similarities, the focus ofinvestigators has been on the commonalities of the humancognitive system (often construed as the role of innatelyavailable structure) in the interpretation of the world. Yetthere are features of the world that are ubiquitous (commonstructure in the environment) that also may be responsible inpart for similarities in cognitive development. Cases include,for example, the fact that objects are always at some distance inrelation to stable landmarks, that objects are always present insome number, etc. Given environmental factors that are more orless always present it is difficult to determine empirically theextent to which the cognitive system begins with structures forprocessing these features or constructs these structures out ofenvironments where these features are generally present.With respect to individual differences, parallel issues arise.Either children begin with differential abilities in dealing withcertain aspects of the world or the environments they encountervary in how likely they are to contain these features. Casesinclude the presentation of spatial information in maps, thepresentation of conventional forms of quantitative informationsuch as number words or use of measurement instruments. I willdiscuss various cases in which.ambiguities arise regarding howbiology and environment may interact in producing the emergingskills of the child in the three domains discussed in thesymposium.