Symposium
Chairs: Deborah A. Kelemen and Tim P. German
Discussant: Karen Wynn
Developmental psychologists increasingly recognize the necessity ofintegrating cross-developmental and cross-species research as a means ofgaining insight into cognitive developmental mechanisms and trajectories.This symposium brings together complementary perspectives from infancy,early preschool and non-human primate research aimed at understandingdevelopments in reasoning about object function and the representation ofartifact categories. Infancy research discussed by our first presenters indicates thatby 14 months, children already possess a general knowledge of artifactfunction, recognizing, for example, that objects with certain featuresafford certain activities (such as being 'drinkable from'). However,results from their generalized imitation procedure indicate that by 19months, infants have begun to display a more detailed knowledge of thefunctions of specific artifact categories. These authors therefore suggestthat infants' first understanding of many artifact functions is at aglobal, abstract, level that becomes increasingly specified and develops,in this culture, in advance of comparable detailed knowledge about animals.Following on from this infancy research, findings by our second presentersraise questions about the necessity of language capacities in structuringour understanding of object function. Through their study of rhesusmacaques and cotton-top tamarins, these researchers conclude thatnon-linguistic non-human animals possess a 'global' understanding of thestructural properties relevant to tool function. Furthermore this knowledgecan be refined in non-human primates if they are given additional exposureto tools. Our last two presentations focus on integrating recent infancyresearch into an understanding of later developments in children'sreasoning about function. Both sets of researchers emphasize the relevanceof early theory of mind abilities to children's developing conception ofartifacts as 'objects designed for purposes'. The first of thesepresentations argues that, during infancy and preschool, children conceiveof artifacts in terms of their general utility rather than in terms oftheir designed function, as adults do. This flexible view of objectfunction renders young children immune to the classic psychological effectof 'functional fixedness'--the phenomenon in which adults' prior knowledgeof an artifact's function blocks their recognition that the object can beused in a novel way to solve a problem. The second of these presentationsexplores the origin of children's bias to broadly view the properties ofboth artifacts and non-living natural kinds in terms of a function. Theauthors examine whether this bias is accounted for by differences betweenyoung children's and adults' conception of function. Based on findings withforced-choice vignettes and behaviorally-active variants of functionalfixedness tasks, these presenters conclude that, by preschool, children'sfunction concept is largely adult-like. They therefore consider alternativeexplanations for children's broad function bias and suggest that its originmay be traced to intention-based explanatory strategies employed ininfancy.
Details of individual items:
paper
We have been using the technique of generalized imitation to assessinfants' knowledge about objects. We model events using little replicas,such as giving a person a drink from a cup. Then we give the infants thecup and two different containers (e.g., a mug and a frying pan) and seewhich they use to imitate the event they watched. In a series ofexperiments testing a variety of artifacts and animals we found that at 14months infants have only a very general understanding of the function ofmany artifacts. For example, they are as apt to imitate the drinking eventby using the frying pan as the mug. Similarly, after watching a plate beingwashed in a sink, they are as apt to imitate this event by washing theplate in a bathtub as in another sink. They also imitate brushing hair witha toothbrush, and hammering with a wrench. It appears that they have somevery general ideas of 'being drinkable from' or 'making hair look nice' butlack many details. We checked to see if the 14 month olds might have learned some ofthe properties we tested in a less meaningful way than the gloss of 'beingdrinkable from' or 'making hair look nice' suggests. They might havelearned more neutral physical descriptions of the activities. For example,hair grooming might mean no more than running any elongated object over thehair. So we conducted a control experiment to show that although theirunderstanding of the modeled activities was overly general it wasnevertheless in the right conceptual ballpark, not merely at the level ofpurely physical description. We tested 14-month-olds on the same propertiesagain, this time using physically similar exemplars from unrelatedcategories as the foils. For example, for washing dishes the foil was a bedwith headboard, footboard, and sides, and for brushing hair the foil was aspoon. Each of these inappropriate objects 'afforded' the modeled activitybut came from a conceptually unrelated domain. Now 14-month-olds weresuccessful, indicating they have some general ideas of functions such ascontainment. By 19 months infants were generally correct on these tasks. Otherexperiments showed that by this age they have also have learned someproperties specific to particular vehicles, such as that gasoline is putinto cars but not into toy wagons. However, they are slower to learn thespecific characteristics associated with various animal kinds, such as thatdogs eat bones but birds do not or that birds sleep in nests but thatrabbits do not. Even at 19 months, choice of appropriate object to imitateanimal activities is still at chance. Not until two years are infantsusually correct. This work indicates that at first infants understand the functionof many objects (or in the case of animals, their behavior) mainly at anabstract, global level. Their understanding only gradually becomes concreteand detailed with experience. In our culture this advance happens earlierfor artifacts than for animals.
paper
Several experiments now provide detailed knowledge about what humanchildren understand about object function. We know much less about suchunderstanding in pre-linguistic human infants and non-linguistic non-humananimals. Our lack of knowledge about non-human species is surprising giventhe fact that work on animal tool use dates back at least 30 years. What donon-human animals understand about the functional interactions betweenobjects? Specifically, what properties do non-human animals attend to whenchoosing between two possible tools or observing interactions betweenobjects? In this presentation, we address these problems by presenting theresults of experiments on two non-human primate species, cotton-toptamarins (Saguinus oedipus oedipus) and rhesus macaques (Macaca mulatta),using two different procedures, a manipulative reaching task and anexpectancy violation task. Results from the manipulative reaching taskreveal that tamarins attend to the functionally relevant properties of apotential tool (e.g., shape, length), while ignoring changes that arefunctionally irrelevant (e.g., color, texture). In the expectancyviolation procedure, both tamarins and rhesus watched a pushing tool act onanother object. Subjects were familiarized to an L-shaped tool pushing agrape across a stage and down a ramp. After familiarization, we variedeither a property that was relevant to the tools function (i.e., its shape)or one that was irrelevant (i.e., its color). We found that subjects inboth species looked longer at changes in the functionally-relevantproperty than changes in the functionally-irrelevant property. Namely,subjects looked longer at the newly shaped tool pushing the grape than thenewly colored tool. We then examined what types of shape change subjectsfound to be relevant to the tool's function. We found that althoughtamarins attended to changes in an object's functional properties (i.e.,the orientation of the tool with respect to the grape), rhesus attended toshape changes alone. Unlike rhesus, the tamarins had substantial experiencemanipulating tools as a result of prior studies in our lab. We proposethat some non-human primates begin with a predisposition to attend to atool's shape and, with sufficient experience, develop a more sophisticatedunderstanding of the features that are functionally-relevant to tool-use.Results are discussed in light of current theories concerning artifactualrepresentations and the importance of language in structuring thisrepresentation.
paper
There is increasing evidence that children begin to understandmental states very early in life. A variety of methods, including lookingpreference, habituation, imitation, gaze following and patterns ofextension of novel words, have suggested that infants are sensitive to thegoals and intentions of others. Some of this work has looked at how infantsand young children understand goal directed activity with objects, i.e.understanding tool-use and simple object functions (see e.g. Mandler'stalk, this symposium). How does this initial knowledge about function impact on laterdevelopment? Colleagues and I have argued that children's intuitions aboutobject function remain embedded in their commonsense psychology through thepreschool years. Specifically, children continue to view objects in termsof quite general functions (e.g. 'what can I do with this?') rather thanacquiring constrained notions of object function (e.g. 'This was designedfor x'). Evidence for this view comes from studies assessing olderpreschool children's intuitions about which of two candidate activities isan object's true function. When choosing between the activity for which anobject was designed and an intentional activity that occurs later, childrenselect the original designed activity less than do adults. In some studies,they select each activity equally often. In control stories, childrenchoose intentional activities over those that occur accidentally. The focus of the current presentation is converging evidence forthe 'agents-goals' position, collected from young preschoolers' andchildren's performance on object based insight problems (e.g. Duncker'sfamous 'candle' problem). Adults exhibit 'functional fixedness' by beingslow, or failing, to use an object for an alternate function in thesolution of a problem, when that object's conventional function has beenpreviously demonstrated. For example, if a box must be used as a platformfor an object rather than to contain objects, then adults are better if thebox is initially presented empty rather than filled. Using a task analogous to Duncker's candle problem, we've shownthat young children are immune to functional fixedness. In one series ofstudies, 5-year-old children were faster than children two years theirsenior in solving a problem involving the use of a toy-box as a platformrather than a container, when the box was presented filled. When thecontainment function was not demonstrated, the reverse age effect wasfound. This result was shown to be replicable, and unaffected by drawingall children's attention to the test object using a verbal label, amanipulation that has attenuated functional fixedness in adults. Theseresults suggest that younger children's flexible notions of object functionrender then immune to functional fixedness. Older children, like adults,can be blocked in reaching problem solutions by more constrained intuitionsabout object function. In further series of studies we have developed a wider range ofobject use insight problems, some suitable for addressing the solution ofsimple insight problems in younger preschoolers and older toddlers. Resultsso far are consistent with the view that young children's initial, fluidnotions of function impact on both their later intuitions about objects andobject-based problem solving.
paper
Question: 'Why are rocks pointy?' Answer: 'So that animals won'tsit on them and smash them.' For most Western adults this is anunacceptable explanation for why a natural object like a rock possessessuch a property. Even for the most geologically uninformed, a moreappropriate response involves a physical process, for example, 'rocks arepointy because material builds over time' or 'because material chippedaway.' Our studies suggest, however, that for young children the oppositeis true: non-living natural object properties exist for a function. Sand isgrainy so that animals can lay eggs in it, and ponds are waveless so thatanimals can bathe safely. Such broad function-based intuitions are presentfrom at least 4- to 5-years of age, extending into early elementary school.But, why do children attribute functions so promiscuously? One possible explanation is that preschool children overapplyfunction-based 'teleological' explanations because they have a differentconcept of function than adults do. For an adult, a function is an activitywith a certain kind of design history but, perhaps, for a child, a functionis any activity that an object can perform. However, research indicates that, even from infancy, children'sunderstanding of function is never entirely unconstrained in this way. Wewill review evidence which increasingly suggests that a conception offunction develops in concert with theory of mind abilities and thatinfants' hypotheses about the purpose of novel objects are constrained byan early sensitivity to the goal-directed behavior of agents. Such sensitivity motivates another possible explanation forchildren's promiscuous attribution of function. Perhaps children applyfunction broadly because their conception of function suggests that anobject is 'for' anything that a individual might intentionally use it toperform, whether designed for that purpose or not. Although, this description may accurately characterize children'sfunction concept during earlier development (see other talks, thissymposium), we will argue that it does not account for the over-applicationof teleological-functional explanation by 4- and 5-year-olds. We willpresent findings which suggest that by this age, children's functionconcept is largely adult-like: rooted in beliefs about the design historyof an object. One source of evidence for this comes from studies which findthat when they are asked to consider novel objects that were designed forone activity but intentionally used for something else, preschoolers andadults agree that the objects are 'for' their designed function. Anothersource is from studies of preschoolers' performance on 'functionalfixedness' tasks. 'Functional fixedness' is the classic phenomenon in whichknowledge of an artifact's function (e.g. a box's containment function)interferes with adults' ability to use it in a novel way whenproblem-solving (e.g. using the box as a platform). In our task,preschoolers design--and therefore 'fix'--the functions of the artifactsthemselves. Preliminary findings suggest that when asked to solve problemsthat involve using the objects in alternative ways (e.g., making a puppetthen using it as a cover), preschoolers representation of the objects'intended function provokes more problem-solving difficulties than incontrol conditions where the problem-solution items have not had theirfunctions explicitly 'fixed'. Further findings from the 'functionalfixedness' research and alternative explanations for children's promiscuousfunctional attributions will be discussed.