Acquiring an understanding of design: evidence from children's insight problem solving

Abstract

The human ability to make tools and use them to solve problems may not be zoologically unique, but it is certainly extraordinary. Yet little is known about the conceptual machinery that makes humans so competent at making and using tools. Do adults and children have concepts specialized for understanding human-made artifacts? If so, are these concepts deployed in attempts to solve novel problems? Here we present new data, derived from problem-solving experiments, which support the following. (i) The structure of the child's concept of artifact function changes profoundly between ages 5 and 7. At age 5, the child's conceptual machinery defines the function of an artifact as any goal a user might have; by age 7, its function is defined by the artifact's typical or intended use. (ii) This conceptual shift has a striking effect on problem-solving performance, i.e. the child's concept of artifact function appears to be deployed in problem solving. (iii) This effect on problem solving is not caused by differences in the amount of knowledge that children have about the typical use of a particular tool; it is mediated by the structure of the child's artifact concept (which organizes and deploys the child's knowledge). In two studies, children between 5 and 7 years of age were matched for their knowledge of what a particular artifact “is for”, and then given a problem that can only be solved if that tool is used for an atypical purpose. All children performed well in a baseline condition. But when they were primed by a demonstration of the artifact's typical function, 5-year-old children solved the problem much faster than 6–7-year-old children. Because all children knew what the tools were for, differences in knowledge alone cannot explain the results. We argue that the older children were slower to solve the problem when the typical function was primed because (i) their artifact concept plays a role in problem solving, and (ii) intended purpose is central to their concept of artifact function, but not to that of the younger children.

Introduction

Using tools to solve novel problems is one of the paradigmatic human cognitive capacities (Pinker, 2002), although it does not distinguish us qualitatively from other animals: ‘tool use’ has been documented in the wild in a number of primates (Hauser, 1997, Tomasello and Call, 1997) and, probably most strikingly, recently in the laboratory in the New Caledonian crow (Weir, Chappell, & Kacelnik, 2002). Nevertheless, the wide variety, complexity and technological sophistication of human tools and artifacts surely make humans the ‘ultimate tool users’. A brief tally of the relative numbers of artifacts versus natural objects in one's immediate environment will likely provide a rapid sense of exactly how pervasive human-made artifacts are (Tomasello, 1999).

Despite our being a tool-using species, the cognitive capacities underlying the acquisition of knowledge about artifacts are not well understood, nor are the capacities that allow us to transfer efficiently this knowledge in solving simple novel problems. Although psychological research on these topics began around 70 years ago (e.g. Matheson, 1931, Richardson, 1932), there has been a striking lack of research devoted to understanding the cognitive basis of tool use and its development in humans (see e.g. Want & Harris, 2001) – especially when compared with the efforts devoted to characterizing and understanding tool use and its development in species that succeed only sporadically, such as capuchins and crows.

Most studies investigating the knowledge that adults and children have about artifacts have focused on intuitions about objects and their functions in categorization tasks, function judgment tasks and word extension tasks (e.g. Gentner, 1978, German and Johnson, 2002, Hall, 1995, Kelemen, 1999, Kemler Nelson, 1999, Landau et al., 1998, Matan, 1995, Matan and Carey, 2001, Rips, 1989). But such tasks may fail to reveal important aspects of artifact knowledge – aspects relevant to problem solving. Indeed, one could argue that ‘core’ conceptual systems (e.g. agency, Leslie, 1994; object mechanics, Spelke, Breinlinger, Macomber, & Jacobson, 1992; and number, Wynn, 1998) ought to promote action and problem solution, not just the mere contemplation of knowledge for its own sake (see e.g. Cosmides and Tooby, 1994, Hood et al., 2000). Herein, we explore the possibility that the conceptual systems that organize knowledge about artifacts do affect human choice and action. More specifically, we investigate the possibility that they play a role in the human capacity to choose tools appropriate for solving novel problems.

Adult reasoning about artifacts appears to reflect the adoption of the ‘design stance’ (Dennett, 1987), an abstract explanatory schema that captures the relationship between features of an entity (e.g. its material, shape and activities) in terms of a coherent organizing notion – the purpose for which its designer created it. Categorization tasks show that adults tend to judge an object's category on the basis of (i) its intended function rather than its appearance (Rips, 1989), and (ii) its designed function rather than its current use (Hall, 1995, Kelemen, 1999, Matan and Carey, 2001). Moreover, adults also judge an object's function on the basis of the original intentions of the designer over other intentional uses and accidental activities (German and Johnson, 2002, Kelemen, 1999).

But what are the developmental origins of the design stance? Recent research and evidence in cognitive development has suggested that developing commonsense understanding of the world is based on what Cosmides and Tooby (2001) call systems of ‘dedicated intelligence’ – rapid learning guided by specialized domains of core knowledge, which allow perception of, attention to, and reasoning about important classes of entities in the children's environment (such as number, object mechanics and agency; see also Spelke, 2000). According to this framework, one possible route by which humans understand artifacts is via mechanisms dedicated to just this process – inference systems that have been selected to represent the category of tools and underlie the capacity for tool use – what Pinker (2002) calls ‘intuitive engineering’ (see also Boyer, 2001).

An alternative possibility for the representation and learning of artifact concepts (albeit also consistent with the core knowledge framework) is the idea that knowledge of artifact function reflects an understanding ‘improvised’ by combining different domains of ‘core knowledge’. First, artifact function representation can be argued to require the capacity to represent and reason about the physical properties of an object and the constraints that those physical properties place on its motion and possible interactions with other objects (e.g. an object's ‘affordances’; Gibson, 1979, Vaina, 1983). Second, because objects can be similar enough in shape and structure to afford exactly the same activities (e.g. an ashtray and a soup bowl), artifact function is also constrained by information about the social agents who create and use those objects to fulfill their goals – information provided by an intuitive psychology (German and Defeyter, 2000, German and Johnson, 2002, Kelemen, 1999).

Under the view that artifact representation might be based on consideration of an object's mechanical structure and the goals of agents using that artifact, the notion of ‘design’ is assumed to emerge and plays a central role in organizing artifact knowledge only later in development (German & Johnson, 2002). While some theorists argue that understanding of design guides reasoning about object functions from about age 4 or even earlier (Kelemen, 1999, Kemler Nelson, 1999, Kemler Nelson et al., 2002, Kemler Nelson et al., 2000), a consensus of evidence derived from a number of tasks suggests the shift occurs somewhat later, around 6 years of age (Gentner, 1978, German and Johnson, 2002, Graham et al., 1999, Landau et al., 1998, Matan, 1995, Matan and Carey, 2001).¹

As noted at the outset, a puzzling gap in the literature on children's conceptual representations of artifacts concerns the deployment of those representations in solving problems that require the use of simple tools. Though the development of means-end problem solving has been studied (Brown, 1990, Sobel, 1939, van Leeuwen et al., 1994), this research had not been linked explicitly to the conceptual representation of artifacts. This link was made explicit by German and Defeyter (2000), who studied the impact of artifact concepts on children's performance on a class of object-use problems made popular by the Gestalt school of psychology (e.g. Duncker, 1945, Maier, 1931). In these tasks, the subject needs to solve a problem using a particular object of known function (variously, a box, a paperclip, a screwdriver, etc.). However, to solve the problem, the tool must be used in an unusual way. For example, in the ‘candle problem’ (see also Adamson, 1952), subjects are presented with a candle, a book of matches and a box of tacks, and asked to fix the candle to a vertical screen. To solve the problem, the tack box must be used as a platform. Adults are far more likely to arrive at this solution – indeed, to find it obvious – when the box is presented without the tacks inside than when the box is presented full of tacks. In other words, priming the box's typical function – containment – makes it more difficult to see that the problem can be solved by using the box in an atypical manner. This phenomenon is called functional fixedness.

Traditional interpretations of functional fixedness propose that accumulated knowledge about the object's regular design function is activated by the demonstration of that function, and this somehow blocks alternative uses which otherwise would come easily to mind, creating an ‘impasse’ (Knoblich, Ohlsson, & Raney, 2001). Both the ‘mental ruts’ hypothesis, which suggests that repeated exploration of knowledge elements in the unsuccessful search for solutions activates incorrect pathways, causing the impasse (Smith, 1995), and the ‘representational change’ hypothesis, which proposes that the initial representation of the problem interacts with prior knowledge in a way that activates knowledge elements that are not helpful in solving the problem (e.g. Kaplan and Simon, 1990, Knoblich et al., 2001), are rooted in this assumption that past knowledge is key to understanding object use when people solve insight problems.

The assumption of the importance of past knowledge is nowhere more pronounced than in discussions of children's learning. In the words of Brown (1989): “one of the more ubiquitous claims concerning young children is that they tend to acquire knowledge in such a way that it is closely tied, restricted or welded to habitual occasions of use” (p. 369). But in explaining functional fixedness in both adults and children, German and Defeyter (2000) have proposed an alternative to explanations based on accumulated experience alone: the emerging design stance hypothesis. On this view, knowledge of a tool's typical function can cause functional fixedness only in solvers whose concept of artifact function embodies the ‘design stance’. The idea is that each individual has an abstract concept of artifact function, and that this concept plays a role in problem solving, organizing knowledge about a tool's possible functions. The hypothesis that functional fixedness is caused by the nature of an abstract concept of artifact function might be difficult to test with adults, given the prevalence of the design stance in adult populations (at least in Europe and America). But it can be tested in children by comparing problem solving by 6- and 7-year-olds – who do reason in accordance with the design stance – to that of 5-year-olds, who appear to lack the relevant design concepts.

German and Defeyter (2000) reasoned that if younger children's representation of artifact function is not based around design, but rather improvised on the basis of representations of an object's mechanical properties on the one hand, and representations of the goals of agents on the other, then an interesting, counterintuitive prediction follows: younger children might be less susceptible than older children to functional fixedness. German and Defeyter (2000) presented 5-, 6- and 7-year-old children with a task analogous to the candle problem. The children's task was to help a puppet reach a high shelf, and the solution was to use a box as a platform (rather than as a container), in order to raise a tower of bricks to the required height. In the key function demonstration condition (i.e. when function was primed), the box was presented in use for its typical function: containment. The bricks and several other inappropriate items (e.g. a coin, pencil eraser, toy car) were presented inside the box. In the baseline condition (function not primed), the box and other items were presented separately. The results showed the following:

• (i) In the baseline condition, the problem was trivially easy for all the children. This demonstrated that, seen merely as a means-end problem, the task was simple.

• (ii) Like adults in the candle problem, 6- and 7-year-olds showed evidence of functional fixedness: they were slower in reaching the solution when the box's typical function was primed than when it was not.

• (iii) In marked contrast, the 5-year-olds showed no evidence of functional fixedness: they solved the problem just as fast when the box's typical function was primed as when it was not. Moreover, the 5-year-olds actually were faster than both 6- and 7-year-old groups in solving the problem when the containment function of the box was first demonstrated.

German and Defeyter (2000) argued that the emergence of fixedness in the older children was not caused by increasing knowledge about the habitual function of specific familiar objects per se, but rather to changes in the way that children reason about artifacts in general – changes in the structure of their artifact concepts. Not only is this supported by prior research suggesting that concepts of artifact function do indeed undergo a shift between age 5 and ages 6–7 (German and Johnson, 2002, Matan, 1995, Matan and Carey, 2001), but it is also consistent with the body of literature showing that by age 5 – and indeed, even earlier – children already have abundant knowledge about everyday object functions (e.g. Abravanel and Gingold, 1985, Gauvain and Greene, 1994, McDonough and Mandler, 1998). This makes it unlikely that the 5-year-olds' lack of functional fixedness was caused by a straightforward lack of knowledge about the typical function of a box.²

Nevertheless, it is still possible that because older and younger children may differ in the extent of their accumulated experience with the habitual functions of familiar objects, older children fail to overcome this ‘force of habit’ when attempting to generate an alternative to the function of containment during the box task of German and Defeyter (2000). Because this initial functional fixedness study was carried out using objects with familiar known functions, as has most often been the case in tool use problem-solving tasks conducted with young children (e.g. objects such as hooks and rakes; Brown, 1989, Sobel, 1939) and indeed invariably in functional fixedness tasks with adult participants too (Birch and Rabinowitz, 1951, Divesta and Walls, 1967, Duncker, 1945, Flavell et al., 1958, Glucksberg and Weisberg, 1966, Yonge, 1966), the precise nature of the interaction between accumulated past knowledge, conceptual structure and problem presentation remains unclear. Remarkably, despite the importance of the idea that ‘negative transfer of past knowledge’ contributes to functional fixedness, there have been no systematic attempts to manipulate the knowledge about object function that participants bring to a functional fixedness problem.

Scholars in cognitive development regularly meet this kind of empirical challenge via the use of arbitrary, novel stimuli such that “children are equated for knowledge by their lack of it” (Brown, 1989, p. 385). In the present case, the use of novel objects in a functional fixedness problem offers widely different predictions on the basis of the ‘accumulated knowledge of habitual use’ versus the ‘emerging design stance’ hypotheses. If children's increasing susceptibility to functional fixedness arises through habitual use of objects for their typical functions, then no such fixedness should result for problems where the objects and functions used are novel. On the other hand, the emerging design stance hypothesis predicts the opposite. Children's emerging sensitivity to ‘design’ in the way their artifact concepts are represented should result in older children interpreting the novel information about the object's function as a core property of the object, just as for familiar artifacts. The priming of the canonical function will then block the generation of alternative uses in the novel object problem too.

There were two aims in the current research. In Experiment 1, the counterintuitive findings of German and Defeyter (2000) were replicated, in an improved object use insight problem with familiar objects. In the second study, the key predictions outlined above were tested in a task involving novel objects matched in terms of mechanical properties (‘affordances’) to the familiar objects of Experiment 1.