Running Head: DEVELOPMENT OF FOLKBIOLOGICAL INDUCTION Cultural and Experiential Differences in the Development of Folk Biological Induction

Carey’s (1985) book on conceptual change and the accompanying argument that children’s biology initially is organized in terms of naïve psychology has sparked a great detail of research and debate. This body of research on children’s biology has, however, been almost exclusively been based on urban, majority culture children in the US or in other industrialized nations. The thesis of this paper is that the folkbiology of urban children may be highly atypical of development because neither the culture nor everyday experience involve as an important component, plants and animals. Two experiments are reported where the participants are urban majority culture children, rural majority culture children, and rural Native American (Menominee) children. The tasks involved category-based induction and judgments about the concept of “alive.” Each group produced a unique profile of development. Only urban children showed evidence for early anthropocentrism, suggesting that the co-mingling of psychology and biology may be a product of an impoverished experience with and exposure to nature. Even the youngest rural children generalized in terms of biological affinity. In addition, all ages of Native American children and the older rural majority culture children (unlike urban children) gave clear evidence of ecological reasoning. These results showing that both culture and expertise (exposure to nature) play a role in the development of folkbiological thought among children underline the precariousness of basing theories of conceptual development solely on studies of urban, majority culture children. Differences in Development of Folk Biological Reasoning 3 Current views of cognition portray concepts as being embedded in theory-like explanatory frameworks (Carey, 1985, Murphy & Medin, 1985; Medin, Lynch, and Solomon, 2000). These framework theories differ in different domains of experience; a framework theory for understanding and predicting the behavior of physical objects necessarily differs from one, which allows us to predict the behavior of sentient beings. Correspondingly, theorists have begun to conceive of conceptual development as a domain-specific process, and have investigated development in core domains such as naïve physics and naïve psychology (Wellman & Gelman, 1999; Wellman & Inagaki, 1997). Another important conceptual domain is that of folkbiology (Medin and Atran, 1999). Folkbiology encompasses how people understand, categorize, and reason about plants and animals. The natural world of plants and animals is pervasive; further, extensive knowledge about living things was critically important in our distant and recent past, and remains so for many people today. A good deal of research has been conducted in the last 15 years on the acquisition of folkbiology, both because of the intrinsic importance of the domain and as a test case for more general ideas about conceptual development. An important idea to emerge from this body of work is that children's understanding of the biological world undergoes a profound shift between ages 4 and 10. Carey (1985, 1995) argues that children's early understanding of plants and animals is anthropocentric. That is, children's understanding of other living things is largely in reference to, or by analogy to, human beings. As a consequence, prototypicality of humans is central to children's conceptions of the biological world. One source of evidence that young children possess an anthropocentric folkbiology comes from a property projection task where children are taught a new fact about a given biological kind (e.g., a dog "has an omentum") and asked whether other kinds (a bird, a fish, a plant) share that property. The rationale is that projection of a novel internal property is an index of the biological affinity between base and target species. By examining patterns of projection, and comparing those patterns to predictions derived from competing theories, we can diagnose which theories children are using to understand the world around them. Carey (1985) taught children properties about one basic-level animal category, then examined projections to other categories differing in similarity to the target category. Participants of ages 4, 6, 10, and adult were taught, that either people, dogs, or bees had a spleen / omentum (e.g., “a green round thing”) inside of them (each subject was only taught on one of the three exemplars). One or two Differences in Development of Folk Biological Reasoning 4 days later participants were asked which of a series of items (human, dog, aardvark, dodo, stinkbug, bee, worm, flower, and some inanimate objects) exhibit each of a series of properties (breathes, has bones, grows, dies), as well as which had the novel property (spleen / omentum). This task relies on the idea that inductive inferences from prototypical members of a category are perceived as stronger than inferences from less central, typical members (Rips, 1975; Osherson et al, 1990). Therefore, an anthropocentric folkbiology makes several predictions: First, if humans are central, prototypical exemplars of living things, then on average projections from humans should be stronger than projections from other living things. Second, an anthropocentric folkbiology might lead to asymmetries in projection. For example, inferences from human to dog might be stronger than from dog to human because of the prototypicality of human (see Osherson, Smith, Wilkie, and Shafir, 1990 for a formal model that predicts asymmetries of inferences between typical and atypical category members). Carey (1985) found that 4-and 6-year-olds exhibited reasoning patterns consistent with an anthropocentric view. Her 4-yr olds readily generalized from humans as a base but they showed little generalization from dogs and almost none from bees as a base. For 6-yr-olds, humans were somewhat privileged bases, as children were more likely to project from humans to other animals (69%) than from dogs to other animals (54%). Asymmetries were evident in comparing human => dog (76%) to dog => human (41%), and human => bee (59%) to bee => human (12%). Thus, 6-year-olds also showed clear evidence of anthropocentric reasoning, particularly in terms of using humans as a privileged inferential base and showing asymmetries in projection from humans versus non-humans. For 10-year-olds and adults, humans are no longer uniquely central, though some effects suggestive of anthropocentrism are still evident in the 10-yr old responses. Carey interpreted these results as supporting a comparison-to-exemplar model of biological reasoning in which the folkbiological gold standard is people. Preschool children use this model for almost all instances of biological inferences. Carey (1985) argues that, “The prototypicality of people plays a much larger role in determining 4-year-olds’ projection of having a spleen than does similarity among animals” (p. 128). Thus, early folkbiology is essentially anthropocentric (See also Johnson & Carey, 1998). More generally, Carey interprets this pattern of reasoning, along with other evidence, as demonstrating that young children possess a qualitatively different understanding of biological phenomena, incommensurate with that of adults. As a consequence, pervasive conceptual change is necessary for children to Differences in Development of Folk Biological Reasoning 5 acquire the adult model in which humans are seen as one animal among many (e.g., Carey, 1999). This anthropocentric view also makes predictions about how children will decide what kinds of things are alive: Similarity to humans rather than a more universal biological criterion should predict live-attributions. A crucial component of any biological understanding is the ability to differentiate living from nonliving things. Although early evidence suggested that young children held beliefs that inanimate objects such as the sun are alive (e.g., Piaget, 1929; Laurendeau and Pinard 1962), more recent evidence suggests that early studies overestimated animistic reasoning. For example, Richards and Siegler (1984) systematically asked children ages 4-11 whether a range of objects (people, animals, plants, vehicles, other inanimate objects) that were described as either being still, being moved, or (where plausible) moving themselves, were alive. Of interest was whether over the entire set of questions, children's responses corresponded to systematic rules. Results showed that children rarely attributed life to vehicles and objects, and never did so systematically. Most younger children systematically attributed life to people and animals, and by around age 8, most children had added plants to the category of living things. Thus the largest developmental shift was not in learning that inanimates are not alive, but rather learning that plants are (see also Carey, 1985; Dolgin & Behrend, 1984; Richards, 1989). Other evidence suggests cultural and experiential differences in patterns of life judgements. Hatano, Siegler, Richards, Inagaki, Stavy & Wax (1993) present data showing that Japanese children may be more liberal in granting life status to objects such as mountains than are US children, and that Israeli children (even older children) are more conservative, often denying even that plants are alive. These findings appear to be tied to cultural beliefs. In some respects the claim that for young children humans are prototypical living things represents a puzzle if not a paradox. Most human cultures draw a sharp distinction between human beings and other animals and one might expect people to be very atypical animals. Johnson, Mervis and Boster (1992) found just that (se also Anglin, 1977). They showed sevenand ten-year-olds, and undergraduates a series of 200 sets of three pictures of mammals and asked them to “point to the two in each set that they thought were most like the same kind of thing.” In general, children and adults showed converging patterns of similarity relationships among mammals. However, adults considered humans more like other mammals than children did. Indeed, when presented with human-nonhuman-nonhuman triads, children almost never Differences in Development of Folk Biological Reasoning 6 paired a human with another animal. This suggests that children do not see humans as especially typical of living things; rather, children may see humans as much more peripheral mammals than adults do. So why the apparent difference between the Johnson et al findings and the Carey results? In the present study we evaluate the idea that the amount and intimacy of children’s contact with plants and animals as well as their cultural background has a critical influence on the development of folkbiological reasoning. Both dimensions are important in explaining adult reasoning patterns (López et al. 1997; Atran, Medin, Ross et al. 1999) and both may be relevant to children’s reasoning as well. Carey’s study participants were from a highly urbanized population. There is evidence that industrialization has led to biological kinds being less psychologically salient than they were a few centuries ago (Wolff, Medin, and Pankratz, 1998). The extent of this “devolution” or loss of contact with nature may vary as a function of culture and setting (Ross 2001). For example, plants and animals are likely to be much less salient to urban folk than they are for rural folk. Considerations such as these lead to the hypothesis that the young children in Carey's (1985) study generalized more from humans than other animals not because humans are prototypical, but because humans are the only animal about which the children had very much knowledge. In our study we examine the degree to which children with different cultural beliefs and levels of regular exposure to plants and animals reflect anthropocentric folkbiological reasoning. The degree to which a shift from an anthropocentric to a biocentric folkbiology is a universal aspect of conceptual development has not been addressed by previous research. To do so requires looking at conceptual development among children that differ in relevant ways from Carey's population (Coley, 2000). It is important to examine the generality of this anthropocentric pattern of reasoning, on at least two grounds. First, as we have just noted anthropocentric folkbiology may reflect a lack of knowledge about the biological world. More precisely, urban children may not know much about living things other than humans; they may be relative folkbiological novices. Indeed, there is evidence suggesting that knowledge has an impact on young children's reliance on humans as a base for reasoning. Inagaki & Hatano (1987, 1991) find that humans serve as a privileged base for property projection, but that this process is constrained by knowledge. For example, properties are not projected from humans to nonhuman organisms when such an inference would contradict children's knowledge of the non-human in question. This account differs from other models of Differences in Development of Folk Biological Reasoning 7 analogy in that (1) rather than searching for most appropriate analogical base, a decision is made on whether humans are appropriate or not, and (2) object-specific knowledge is used, not to choose an appropriate analogical base, but rather to (a) judge the feasibility of the alreadypredicted behavior, and (b) compute the organism’s similarity to humans. Moreover, this account differs from Carey’s in that humans constitute a privileged analogical base because of children’s relatively rich knowledge about humans, not because of the centrality of humans in children’s biological theories. Inagaki (1990) presented evidence that knowledge does influence children’s use of biological analogy. She compared kindergartners who raised goldfish with their counterparts who did not raise goldfish. All children were asked questions about observable and nonobservable properties of goldfish, asked to reason about goldfish in novel situations, and asked to reason about a novel aquatic animal (a frog) in similar situations. Children who were raising goldfish possessed more knowledge about both observable and unobservable attributes of goldfish. They were also more likely to make reasonable predictions about the behavior of goldfish in novel situations. Although there were no differences in number of reasonable predictions about frogs between the goldfish-raising and non-goldfish-raising groups, the goldfish-raisers were more likely to make reasonable predictions about the behavior of frogs accompanied by comprehensible justifications. Most importantly, while both groups tended to analogize from humans to frogs when answering questions about frogs, the goldfish-raisers were more likely to analogize from goldfish to frogs when answering the same questions. This suggests that knowledge of goldfish enabled children who were raising goldfish to use goldfish as an analogical base in a way that children who were not goldfish raisers could not. A second possibility is that an anthropocentric folkbiology may reflect cultural assumptions about relations between humans and nature. Popular films (e.g. Disney productions) and children’s books often anthropomorphize animals by giving them distinctly human characteristics such as speech, driving vehicles and the like. Religions differ in the extent to which they emphasize the uniqueness of humans versus the rest of nature. In a culture where humans are perceived as an integral part of nature, such as traditional Native American culture, children might be less likely to make anthropocentric construals. In this study, we examine how differences in experience and/or culture might lead to differences in reasoning about plants and animals. Following Carey, we teach children novel Differences in Development of Folk Biological Reasoning 8 properties about humans, wolves, bees, goldenrod, and water, and then see whether they are willing to project these new properties to an array of animals, plants, and nonliving objects. We also ask children whether each member of this array is alive, and why. Of central importance is the question of comparative development; how do experience and cultural beliefs about nature impact conceptual development? To address this question we examine children from three distinct populations; Native American children from the Menominee Indian Tribe of Wisconsin, Rural Majority Culture children from the neighboring town of Shawano, Wisconsin, and Urban children from East Boston, Massachusetts. For a number of reasons, the Menonimee population is of particular interest for this study. First, on the traditional Native American view, humans are an integral part of the natural world (Bierhorst, 1994; Suzuki & Knudtson, 1992). This contrasts sharply with the Western view in which humans are seen as distinctly apart from nature. Second, traditional folkbiological knowledge is especially salient to the Menominee. Unlike most woodland tribes, the Menominee reservation occupies (a small fraction of) their traditional range; thus, traditional knowledge of local plant and animal species is still very relevant today. Overall, sustainable coexistence with nature is a strong value among members of this population (Hall & Pecore 1995). Children are introduced to fishing & hunting at a very early age and in general have a very high degree of contact with plants and animals. The Shawano population is a useful comparison because the town is immediately adjacent to the Menominee Reservation. Children in Shawano grow up in the more or less the same physical environment, are introduced to fishing & hunting at an early age and also have a great deal of first hand experience with the natural world. They differ from the Menominee in terms of the cultural and religious significance of their natural surroundings. Finally, the urban population can be said to share some basic cultural beliefs about the relation between humans and the natural world with the Shawano population, but differs from both Wisconsin populations in that Boston kids have very little firsthand interaction with nature in their daily lives. Examining these populations allows us to examine the pervasiveness of anthropocentric origins of folkbiology, and to begin to "triangulate" with respect to possible causes of conceptual differences (Bailenson, Shum, Atran, Medin, and Coley, in press; see also Coley, 2000). To the degree that the two Wisconsin populations are similar, experience is implicated in shaping folkbiological beliefs. To the degree that the Shawano children resemble the Boston children Differences in Development of Folk Biological Reasoning 9 rather than the Menominee children, a role of cultural beliefs about nature is suggested. Distinct patterns among the three populations might suggest a combination of these factors in shaping conceptual development. Commonalities among the groups would suggest candidates for universals in development. Although these comparisons are by necessity imperfect, they allow us to begin to address the crucial question of how cultural and experiential context act to influence conceptual development. In parallel with the present studies our research team has also examined inductive reasoning about biological kinds in Yukatek Maya children in southcentral Quintana Roo, Mexico (Atran, Medin, Lynch, Vapnarsky, Ucan Ek’, and Sousa, 2001). There we found that even the youngest children tested (4-5 yr.-olds) showed no evidence of anthropocentrism; they generalized readily from both humans and other animals as a function of biological relatedness. Young girls showed less differentiation than boys when the peccary was the base for induction, a pattern consistent with an effect of experience or familiarity. Boys go with their fathers into the forest at an early age and, therefore, are much more familiar with the peccary than are girls. Children generalize more from humans to the sun than from any other base to the sun. This may reflect the special role of the sun in Maya cosmology, though Yukatek adults do not show this pattern. The Atran et al, 2001 studies support the idea that some combination of culture and experience affects children’s biological inductions. The present study will help tease apart the contributions of these two variables. Our study differed from Carey’s in that we included two different kinds of inductive bases, goldenrod and water. We added goldenrod so that we could examine generalization both from animals to plants and from plants to animals. We had reason to believe that Menominee children would have a broad view of living kinds that includes not only plants but also natural entities such a rocks. Our original motivation for including water as a base was to see if ecological relations might play some role in children’s inductions. Previously we had found that adults knowledgeable about biology often rely on ecological reasoning strategies (Lopez, et al, 1997, Proffitt, Coley, and Medin, 2000) and we were interested in whether and when it might appear in children’s reasoning. As it turned out, however, when water is a base children use a wide variety of strategies and it is difficult to draw any clear conclusions. To reduce the complexity of an already complex design, we do not present the results for water as a base in this paper. Nonetheless, we were able to educe evidence for ecological reasoning from other bases. Differences in Development of Folk Biological Reasoning 10