A New Agenda for Cognitive Development

My goal in writing this book is to change the agenda of the field of cognitive development. In particular, I want to promote greater attention to the question that I believe is inherently at the core of the field: How do changes in children’s thinking occur? Focusing on change may not sound like a radical departure from current practice, but I believe it is. It will require reformulation of our basic assumptions about children’s thinking, the kinds of questions we ask about it, our methods for studying it, the mechanisms we propose to explain it, and the basic metaphors that underlie our thinking about it. That modifications of all of these types are being proposed as a package is no accident. Just as existing approaches have directed our attention away from the change process, so may new ones lead us to focus squarely on it. This concluding chapter summarizes the kinds of changes in assumptions, questions, methods, mechanisms, and metaphors that I think are needed. My initial decision to write this book was motivated by a growing discomfort with the large gap between the inherent mission of the field—to understand changes in children’s thinking—and most of what we actually have been studying. As I thought about the problem, I came to the conclusion that existing assumptions, methods, and theories acted in a mutually supportive way to make what we typically do seem essential, and to make doing otherwise—that is, studying change directly—seem impossible. Even approaches that proclaimed themselves to be radical departures from traditional theories maintained many fundamental assumptions of those theories. An increasing body of empirical evidence, however, indicates that some of the assumptions are wrong and that the way in which they are wrong has led us to ignore fundamental aspects of development. In this section, I describe prevailing assumptions regarding variability, choice, and change, and propose alternatives that seem more consistent with empirical data and more useful for increasing our understanding of how changes occur.

Formal Models of Strategy Choice

Pittsburgh has many beautiful old houses, most of which have thick plaster walls. The walls are good in the sense that they effectively reduce noise coming from other rooms, but bad in the sense that they sometimes force me to interact with plasterers. My experience with plasterers has not been fun. The ones who have come to do repairs on my house haven’t put drop cloths below where they worked, have resisted or ignored requests not to smoke in the house, and have flicked cigarette ashes wherever they went. Several of them smelled bad and had alcohol on their breath, even at 8:00 A.M. On the other hand, they were really good at plastering. They fixed cracks and holes quickly and without apparent difficulty. The finished product, once painted, matched the surrounding wall or ceiling remarkably well. My own attempts at plastering have been much less successful. These attempts have taught me quite a bit, but unfortunately not how to do the job. I have learned that plastering requires many good strategy choices. Home plastering kits indicate how much plaster powder and water to mix together, and roughly when to apply the mixture, but that is only the beginning of the problem. When exactly has the plaster jelled sufficiently to apply it? If it’s too watery, it will run; if it’s too thick, it will be wavy, and soon will crack. The instructions also do not illustrate the wrist movements for applying the plaster. The package directs the reader to apply the plaster with “flexible,” “supple,” and “smooth” movements, but what does that mean? Despite considerable time spent thinking about how to plaster, and some time spent watching plasterers, my decisions invariably lead to wavy, cracked surfaces resembling those of an artistically challenged kindergaitner. I occasionally have asked the plasterers who came to my house how they know when the plaster is ready and how it should be applied. Their replies: “You just look at it.” “You can see when it’s ready.” “You put it on smoothly.” True enough, I’m sure, but not helpful.

Whose Children Are We Talking About?

My children have never looked much like those described in most theories of cognitive development. I don’t mean that they are generally deviant or that they perform abnormally on conservation or class inclusion tasks. They generally seem more or less normal, and on the few occasions when I have presented tasks from the cognitive developmental literature, they have acted pretty much like the children described in the articles. Where my children are altogether different from the theoretical descriptions is in the variability of their thinking. Cognitive developmental theories generally depict age and thought as proceeding in a 1:1 relation. At an early age, children think in one way; at a later age, they think in another way; at a still later age, they think in a third way. Such descriptions are so pervasive that they begin to feel like reality. Young children are said to form thematic concepts; somewhat older ones to form chain concepts; yet older ones to form true concepts. The reasoning of young children is said to be preoperational; that of somewhat older ones concrete operational; that of yet older ones formal operational. Young children are said to have one theory of mind; somewhat older ones a different, more inclusive theory; yet older ones a more advanced theory still. The story is the same with characterizations of performance on specific tasks. In descriptions of the development of the concept of living things, 3- and 4-yearolds are said to think that anything that moves is alive, 5- to 8-year-olds that animals—and only animals—are alive, and older children that plants as well as animals are alive. In descriptions of the development of addition skill, kindergartners are said to count from one; first through third graders to count from the larger addend; fourth graders and older children to retrieve answers from memory. In descriptions of the development of serial recall strategies, 5-year-olds are said not to rehearse; 8-year-olds to rehearse in a simple way; 11-year-olds to rehearse in a more elaborate way. My children’s thinking has never looked as neat and clean as these 1:1 characterizations of the relation between age and thought.

How Children Generate New Ways of Thinking

Cognitive variability ultimately depends on people’s constructing novel ways of thinking even when we already possess reasonable alternatives. If we did not supplement existing approaches with new ones, there would be no variation in the approaches we use to solve a single problem, nor any need to choose adaptively among the alternatives. But how do we generate new approaches, and what leads us to do so when existing approaches already are adequate? This is the focus of the present chapter. Despite the obvious importance of constructing new ways of thinking, we know little about how the process occurs. For every study focusing on how children generate new approaches, there must be a hundred that focus on how the thinking of different-age children differs. Both conceptual and methodological considerations have contributed to this imbalance. Conceptually, the staircase approaches that have been dominant for the last 30 years have depicted generation of new ways of thinking as a rare event, too rare to be studied systematically. The outcomes of the generative process could be observed, as in the studies of age-related differences in reasoning, but not the generative process itself. Methodologically, standard developmental approaches, whether cross-sectional or longitudinal, are not well suited to studying the process of change. Developmental psychologists’ reliance on such approaches in part reflects the conceptual blinders imposed by the staircase conceptions of development. If generation of new approaches is a rare event, then these methods might be the best possible. Expedience also plays a role; cross-sectional methods in particular are cheaper and easier to use than methods more specifically designed to get at the process of change. Habit also plays a role; the standard methods are so prevalent, and have been for so long, that they almost inevitably are the first methods that come to mind when thinking about how to study a developmental issue. However, they are not the only methods for studying development, nor the ones best suited to studying change processes. An alternative approach that seems especially promising for studying change is the microgenetic method.

Cognitive Variability: The Ubiquity of Multiplicity

No one doubts that immense variability exists at the neural level. Even when the identical stimulus is presented repeatedly within a single experimental session, the response of an individual neuron varies from trial to trial. Similarly, with lowlevel cognitive processes such as association, there is no disagreement concerning the existence of competing units. Models of associative memory, both symbolic (e.g., Gillilund & Shiffrin, 1984) and subsymbolic (e.g., Seidenberg & McClelland, 1990), are predicated on the assumptions that stimuli have multiple associations and that these varying associations influence the way in which we remember. Higher level cognition, however, has been treated differently. Many models are universalist: Everyone is depicted as proceeding in the same way when relevant stimuli are presented. Other models are comparative; they hypothesize different ways of thinking among groups defined on the basis of such characteristics as age, expertise, or aptitudes, but hypothesize a single consistent kind of reasoning within each group. Thus, 8-year-olds might be depicted as performing in one way and 5-year-olds in another, experts in one way and novices in another, people with high spatial ability in one way and those with low spatial ability in another, and so on. The finest differentiations that are typically made within these comparative approaches examine individual differences within people of a single age; for example, reflective 8-year-olds are described as taking a long time but answering accurately on the Matching Familar Figures Test, and impulsive 8-year-olds are described as answering more quickly but less accurately (Kogan, 1983). The main purpose of this chapter is to summarize the rapidly growing body of research suggesting that variability is actually a pervasive reality in high-level, as well as low-level, cognition. To place this work in context, however, it seems useful first to briefly consider some prominent examples of universalist and comparative models of cognition and then to consider why they might be proposed and widely accepted even if thinking is far more variable than they depict it as being. A great deal of cognitive research has been devoted to identifying the processing approach that people use on a particular task. This universalist approach has led to many influential models and theories.

The Adaptivity of Multiplicity

By itself, cognitive variability would be an interesting curio, but no more. What makes it important is the potential it offers for adapting to task and situational demands. Realizing this potential, however, requires wise choices among the available alternatives. If children know two strategies—one faster, the other more accurate—they will benefit only if they choose the faster strategy when speed is most important and choose the more accurate when accuracy is. Choosing randomly will yield worse performance than always using the approach that on average yields the best outcome. Thus, the higher the quality of choices among alternative approaches, the greater the benefits of cognitive variability. How well do children choose? The conclusions of developmental psychologists have tended to emphasize the negative. Those interested in metacognition have focused on findings that children who have been taught new strategies often do not use them when they later are free to choose (e.g, Ghatala, Levin, Pressley, & Goodwin, 1986; Keeney, Cannizo, & Flavell, 1967; Paris & Lindauer, 1982; Williams & Goulet, 1975). Those interested in decision making have focused on situations in which children choose alternatives with lower expected values over ones with higher values (Byrnes & McClenny, 1994; Klayman, 1985). Those interested in planning have focused on situations in which children who have heard the virtues of planning extolled nonetheless fail to plan (Friedman, Scholnick, & Cocking, 1987). Those interested in scientific reasoning have emphasized cases in which children who have discovered advanced experimentation strategies nonetheless continue to choose less advanced ones (Kuhn, Amsel, & O’Laughlin 1988). These and related findings have led to generally negative conclusions about children’s ability to choose wisely among alternative ways of thinking. To cite two examples: . . . Children do not monitor well and often fail to make appropriate executive decisions.

Strategic Development: Trudging up the Staircase or Swimming with the Tide?

As described in the previous chapter, children’s thinking is highly variable. The present chapter focuses on implications of this variability for understanding development. Thinking about a pair of visual metaphors may help facilitate recognition of these implications. The first, which I believe underlies most depictions of development, is the staircase metaphor. The second, which I believe offers a superior alternative, is the overlapping waves metaphor. Cognitive developmentalists have often phrased their models in terms that suggest that children of a given age think about a given task in a single way. W-year-olds are said to have a particular mental structure, a particular processing limit, a particular theory, strategy, or rule that gives rise to a single type of behavior. Change involves a substitution of one mental entity (and accompanying behavior) for another. The basic conceptualization that seems to underlie these models is aptly captured in the title of Robbie Case’s (1992) recent book The Mind’s Staircase. The visual metaphor that this title evokes is, I believe, central to most cognitive-developmental treatments of change: Children are depicted as thinking in a given way for an extended period of time (a tread on the staircase); then their thinking undergoes a sudden, vertical shift (a riser on the staircase); then they think in a different, higher way for another extended period of time (the next tread); and so on. This view of development is most closely identified with Piagetian and neo-Piagetian approaches, such as those of Piaget and Case. Thus, as shown in Figure 4.1, we see development depicted within Piaget’s theory as involving sensorimotor activities from birth to about 2 years; preoperational thinking from 2 to 7 years; concrete operational thinking from 7 to 12 years; and formal operational thinking from 12 years onward. Within Case’s theory, we see thinking depicted as advancing from the sensorimotor level between birth and 18 months, to the relational level from 18 months to 5 years, to the dimensional level between 5 and 11 years, and to the formal level at age 11 and beyond. Although this view of development is associated with the Piagetian and neo-Piagetian traditions, it is far from unique to them.

Evolution and Cognitive Development

The central argument of this chapter is that concepts that have helped biologists understand the evolution of species can also help cognitive developmentalists understand the growth of children’s thinking. This argument rests on four main points, corresponding to the four main sections of the chapter. The first point is that current theories of cognitive development are limited by lack of understanding of change mechanisms. The second point is that the problem of explaining the evolution of species has important commonalities with the problem of explaining changes in children’s thinking. The third point is that explanations emphasizing variation, self-regulation, adaptive change, and inheritance have proved useful in understanding biological evolution. The fourth point is that these same concepts are implicitly present in a number of the most promising mechanisms that have been proposed to account for cognitive-developmental change, and that it may be generally useful to try to understand cognitive-developmental change in terms of them. The rest of this chapter elaborates these themes. At present, there is no dominant theory of cognitive development. The limitations of the major theories in the area—Piagetian, neo-Piagetian, Vygotskian, information processing, social learning, ethological, and neo-nativistic—are sufficiently large and apparent that none of them can claim the adherence of anything like a majority of investigators. In all likelihood, the greatest number of developmentalists see themselves as eclectic, borrowing concepts from many theories, but not being entirely comfortable with any one of them. After reviewing current theories of development, Miller (1993) concluded that they share three weaknesses: overly narrow scope, uncertain ecological validity, and inadequate accounts of developmental mechanisms. Of these, the inadequate accounts of developmental mechanisms seems the fundamental limitation. A better understanding of mechanisms would contribute to overcoming the other two problems; it would expand the range of phenomena that could be understood as products of the same underlying mechanisms, and these would likely include everyday as well as laboratory phenomena. The converse does not seem likely: The vast accumulation of data we already possess has generated only limited progress in understanding mechanisms, and it is difficult to see how a broader range of data and/or more ecologically valid data would change the situation much.