Elsevier

Cognition

Volume 168, November 2017, Pages 335-343
Cognition

Original Articles
Young children’s tool innovation across culture: Affordance visibility matters

https://doi.org/10.1016/j.cognition.2017.07.015Get rights and content

Abstract

Young children typically demonstrate low rates of tool innovation. However, previous studies have limited children’s performance by presenting tools with opaque affordances. In an attempt to scaffold children’s understanding of what constitutes an appropriate tool within an innovation task we compared tools in which the focal affordance was visible to those in which it was opaque. To evaluate possible cultural specificity, data collection was undertaken in a Western urban population and a remote Indigenous community. As expected affordance visibility altered innovation rates: young children were more likely to innovate on a tool that had visible affordances than one with concealed affordances. Furthermore, innovation rates were higher than those reported in previous innovation studies. Cultural background did not affect children’s rates of tool innovation. It is suggested that new methods for testing tool innovation in children must be developed in order to broaden our knowledge of young children’s tool innovation capabilities.

Introduction

The extent to which humans innovate with tools remains unparalleled within the animal kingdom (Carr et al., 2016, Vaesen, 2012). Yet the capacity for tool innovation appears curiously absent in young children, with multiple studies showing that prior to 8 years of age children struggle to innovate even simple tools on their own (Beck et al., 2011, Beck et al., 2016, Cutting, 2013, Cutting et al., 2014, Nielsen, 2013). This is curious, as from a young age children are adept tool users (Brown, 1990, Connolly and Dalgleish, 1989, Harris, 2005). However, previous studies may have limited children’s performance by presenting tools with opaque affordances. In addition, the vast majority of testing to date has been conducted using the same methodology, and tested almost exclusively children from Western cultural backgrounds (Nielsen, Tomaselli, Mushin, & Whiten, 2014). These factors may individually or in combination lead to apparent tool innovation failure that may not accurately portray children’s true capacities.

Children are driven to explore and utilize the material world around them (Bakeman et al., 1990, Bock, 2005, Gaskins, 2000, Kaye, 1982, Keller et al., 2009, Little et al., 2016, Piaget and Cook, 1952, Rogoff et al., 1993). By the age of four months, infants from Western and traditional societies demonstrate a sustained interest in objects, and by 8–11 months begin to engage in relational play with objects (Belsky and Most, 1981, Bjorklund and Gardiner, 2011, Bourgeois et al., 2005, Konner, 1976). This interest persists well into the early childhood years, manifesting as object play, construction and manipulation (Bakeman et al., 1990, Belsky and Most, 1981, Bock and Johnson, 2004, Little et al., 2016, Smith and Simon, 1984), as children examine the causal relationships existing between objects and the environment (Bjorklund and Gardiner, 2011, Lockman, 2000, Pepler and Rubin, 1982, Piaget and Cook, 1952). At the age of nine months children begin to use tools to reach for objects far away from them (Willatts, 1984), and by two years they can competently use tools such as spoons and rakes (Brown, 1990, Connolly and Dalgleish, 1989, Harris, 2005, McCarty et al., 2001). They can even invent simple tool-use behaviors independently by three years (Reindl, Beck, Apperly, & Tennie, 2016). Young children are also capable of tool manufacture: constructing or modifying tools after watching an adult manipulate relevant materials (Barr and Hayne, 1999, Bauer et al., 1995, Beck et al., 2011, Cutting et al., 2011). While tool manufacture occurs following observation or instruction on how to make the ideal tool (Cutting et al., 2011, Shumaker et al., 2011), tool innovation necessitates the construction of a novel tool that is designed by the individual without previously witnessing a demonstration of the means to do so (Cutting et al., 2011). This is a cognitively demanding feat: first the child must generate an ideal tool shape that might solve a task, then they must develop an action plan for creating that ideal tool shape, and finally execute that to an adequate degree to ensure success. It is perhaps unsurprising, then, that children of 4–5 years of age struggle to innovate new tools (Beck et al., 2011, Cutting, 2013, Cutting et al., 2014).

However, by this age children demonstrate developing capabilities in means-end reasoning, working memory, inhibitory control and causal understanding, which are purported to be involved in such multi-step problem solving (Bechtel et al., 2013, Brown, 1990, Chappell et al., 2013, Chappell et al., 2015, Gardiner et al., 2012, Garon et al., 2008, Miyake et al., 2000, Pauen and Bechtel-Kuehne, 2016, Pauen and Wilkening, 1997, Reader et al., 2016; although see Beck et al. (2016) for a lack of relationship between tool innovation and executive function). They have an appreciation of affordances: the relation between an object and an actor, and object and the environment, which provides the actor with an opportunity to perform an action, should they recognise it (Gibson, 1969, Gibson, 1979, Norman, 2013). This begins in infancy with an exploration of object properties such as pliability, flexibility and rigidity (Bourgeois et al., 2005, Fontenelle et al., 2007, Geary, 2005), and progresses to investigations into object relations between form and function in the second year (Bjorklund and Gardiner, 2011, Brown, 1990, Madole et al., 1993, Pauen and Bechtel-Kuehne, 2016). In this way children learn that an object’s form affords action: a spoon affords scooping, and a hook affords pulling (Bjorklund and Gardiner, 2011, Gibson, 1969). Given the sophisticated cognitive toolkit young children are developing, it is reasonable to expect them to be better at tool innovation, yet they appear not to be.

To date, almost all studies examining children’s tool innovation have employed the same basic methodology. The task, which was first administered to New Caledonian crows (Weir, Chappell, & Kacelnik, 2002), involves retrieving a bucket and reward from a long, vertical tube using some form of pliable material. For children, the reward consists of a toy and sticker, which are placed into the bucket, and lowered to the base of the narrow tube. Children are presented with a straight pipecleaner and some distractor items (e.g., a string and some match sticks), and told that these things might help them in retrieving the toy from the tube. Children are then given one minute to retrieve the toy. In order to be successful on the task, children must innovate a novel tool from the materials provided. Without seeing a demonstration of how to do so, they must select the straight pipecleaner and bend its end into a hook-shape, so that it may be placed down the tube and hooked onto the bucket’s handle to lift it up.

Young children find this task extremely challenging: Across a number of studies, only 8–20% of 4–5 year-olds spontaneously make a hook with the pipecleaner (Beck et al., 2011, Chappell et al., 2013, Cutting et al., 2014; although see Sheridan, Konopasky, Kirkwood, and Defeyter (2016) for performance of 44% in 4–5 year-olds). It is only at about 8–9 years of age that 60–65% of children innovate the ideal hooked tool (Beck et al., 2011). When compared with high innovation rates of over 90% in adult samples, it appears that young children are particularly poor at innovating in this task.

What, then, might make this task so difficult for young children? One reason may be its “ill-structured” nature (Chappell et al., 2013). In ill-structured problems, key information necessary for the successful solving of the problem is omitted from the available stimuli (Goel and Grafman, 2000, Wood, 1983). This information must therefore be internally generated by the individual in order for the task to be solved. For example, in the pipecleaner task previously described, children are provided with information about the starting material state (use a pipecleaner, string or matchstick), and the goal state (retrieve the bucket from the tube), but no information is given about how the starting materials might be transformed in order to successfully achieve this end. Instead, the child must independently determine two things: an ideal tool shape to use on the task (a hooked tool), and a strategy on how to construct that shape from the available materials (bend the pipecleaner; Bongers et al., 2003, Cox and Smitsman, 2006).

Consequently, one reason why children may fail to generate the ideal tool shape is because they may not detect the appropriate affordance existing within the material. There is much evidence to show that perceptual information incongruent with the causal properties of a tool will lower overall tool performance (Bates et al., 1980, Gardiner et al., 2012, Gentner and Markman, 1997, Pierce and Gholson, 1994, Rattermann and Gentner, 1998, Winner et al., 1976). A hooked pipecleaner has a “visible” affordance: its ability to complete the action of ‘hooking’ onto the bucket is perceptually obvious. In contrast, in the classic tube problem, the straight pipecleaner offered has a “hidden” affordance: although it has the potential to be bent into a hook, this cannot be perceived in its current state. By providing a hooked pipecleaner, children are given clear information about how the tool might effectively be used to achieve the goal of retrieving the bucket. The straight pipecleaner, however, could have any number of uses or the potential for multiple transformations within the task, and success relies on the child arriving on this hook shape on his or her own in order for it to be used effectively.

Similarly, children perform best at tool-use tasks when the causal link between a tool’s form and its function is highlighted (Bechtel et al., 2013, Gardiner et al., 2012, Goswami and Brown, 1990, Pierce and Gholson, 1994, Winner et al., 1976). Indeed, children’s success on the pipecleaner task elevates if they are given an indication of the ideal tool shape required. Beck et al. (2011) gave children the choice between using a hooked pipecleaner or a straight pipecleaner, and children reliably selected the hooked pipecleaner and used it on the task. This suggests that children can recognise a hook-shape as providing the necessary affordance needed to solve the task, but that they struggle to generate this tool shape on their own.

Alternatively, children might be able to generate the idea of a hooked tool, but struggle to develop an action plan that will transform the straight pipecleaner into that ideal tool (Bjorklund & Gardiner, 2011). Indeed, children will readily copy an adult’s demonstration of how to make a hook – once they see how to bend the pipecleaner’s end upwards, they copy this action and swiftly apply it to the tube problem (Beck et al., 2011, Cutting et al., 2011). This suggests again that children are able to recognise the value of a hooked tool and can readily map the action plan they observed onto their physical materials to create an adequate tool themselves.

Although such scaffolding procedures are valuable in verifying some of the cognitions that underlie children’s tool use, by providing a hooked tool template, they also remove the ‘innovative’ element of the task. These studies have reduced the tube problem from one requiring tool innovation, in which no example of an ideal tool is provided, to one requiring tool manufacture (where a template tool is constructed for the child to copy) or tool use (where the appropriate tool must be selected from an array). It is still unknown whether adding information about the ideal tool shape needed without providing an example of the exemplar tool might see equal improvement in children’s performance on the task.

The current study thus aimed to examine whether providing a pipecleaner that had its hooked affordance visible, but required innovation in another form, would see children’s performance improve on the tube problem. We provided a hooked pipecleaner that had the non-hook end curled over and hence required unbending in order to create the ideal tool. Children thus needed to innovate on the non-hook end of the tool to make it long and straight. This was compared to the performance of children who received the straight pipecleaner as per the classic task, which required one end to be bent into a hook.1 It was hypothesised that children provided with the focal affordance of the target tool (the hook shape) would select this tool more often against a distractor and correctly innovate on the tool at higher rates than children for whom the affordances remained invisible. Providing visual information would reduce the cognitive load inherent in the task, because children would only be required to recognise, rather than generate, the appropriate affordance (and therefore function) of the tool for the task.

Further, calls remain strong for data collection in psychology to move away from reliance on homogenous samples, and specifically those that are Westernised, Educated, Industrialised, Rich, and Democratic (WEIRD; Henrich et al., 2010, Legare and Harris, 2016, Legare and Nielsen, 2015, Nielsen and Haun, 2016, Rowley and Camacho, 2015). This issue is particularly pertinent here with only one study to date examining a non-WEIRD sample, finding poor tool innovation in Southern African Bushman children similar to that of Western children (Nielsen et al., 2014). However, a child’s detection or interpretation of object affordances will be influenced by how they see others interacting with similar objects, and so may be culturally defined (Bakeman et al., 1990, Flynn, 2008, Little et al., 2016, Tennie et al., 2009, Tomasello and Call, 1997, Whiten and Flynn, 2010). The extent to which children’s poor innovation capabilities are culturally-dependent or biologically universal thus remains largely uncharted. We therefore undertook data collection in two distinct cultural samples – children living in a typical WEIRD city and children living in a remote, Indigenous Australian community. Following recent approaches (Little et al., 2016), our goal here was not to emphasize the dichotomy between Western and Non-Western populations but rather to enable better articulation of the universality of young children’s tool innovation abilities. Hence no hypotheses regarding potential differences between these two communities were generated.

Section snippets

Participants

Thirty Indigenous Australian children (16 male, 14 female) aged between 3 and 5 years (M = 4 years 3 months, range = 3 years 2 months to 5 years 10 months) participated in this experiment. Four additional children were tested, but their data was excluded due to excessive shyness (N = 3) or recording failure (N = 1). These children were residents of the Borroloola and Robinson River Aboriginal communities in Northern Australia. Borroloola is a remote town of roughly 1500 inhabitants, with a predominantly

Results

Chi-square tests were employed for all statistical comparisons between the Hook Visible and Hook Not Visible conditions. In comparisons with low expected cell frequencies, Fisher’s exact tests were run and are reported instead. Exact McNemar tests were used to compare between children’s use of each material, and binomial tests were used to assess the frequency of use in the correct orientation (hook-end down) against chance levels. Chance here was defined as 50% because there were only two

Discussion

Previous research has highlighted the difficulties young children experience when innovating novel tools. While children are extremely good at copying the tool-making actions they see (Beck et al., 2011, Cutting et al., 2011), or selecting adequate tools for a task (Beck et al., 2011), they struggle to design and make tools on their own (Beck et al., 2011, Beck et al., 2016, Cutting, 2013, Cutting et al., 2014). The current study sought to examine whether young children could perform better on

Acknowledgements

We thank the Mabunji Aboriginal Resource Centre, and the children and staff at Wunala Creche and Borroloola School for their help in data collection in the Northern Territory. We thank the Early Cognitive Development Centre (ECDC) at the University of Queensland and the Centenary Christian Kindergarten (CCK) and their staff, parents and children for their help in data collection in Brisbane. We thank Matti Wilks for her assistance in data collection and Mitchell Green for his assistance in

References (80)

  • U. Goswami et al.

    Melting chocolate and melting snowmen: Analogical reasoning and causal relations

    Cognition

    (1990)
  • C.H. Legare et al.

    Imitation and innovation: The dual engines of cultural learning

    Trends in Cognitive Sciences

    (2015)
  • K.L. Madole et al.

    Developmental changes in infants' attention to function and form-function correlations

    Cognitive Development

    (1993)
  • A. Miyake et al.

    The unity and diversity of executive functions and their contributions to complex “frontal lobe” tasks: A latent variable analysis

    Cognitive Psychology

    (2000)
  • M. Nielsen et al.

    Exploring tool innovation: A comparison of Western and Bushman children

    Journal of Experimental Child Psychology

    (2014)
  • S. Pauen et al.

    Children's analogical reasoning about natural phenomena

    Journal of Experimental Child Psychology

    (1997)
  • M.J. Rattermann et al.

    More evidence for a relational shift in the development of analogy: Children's performance on a causal-mapping task

    Cognitive Development

    (1998)
  • G. Vingerhoets et al.

    Conceptual and physical object qualities contribute differently to motor affordances

    Brain and Cognition

    (2009)
  • R. Bakeman et al.

    !Kung infancy: The social context of object exploration

    Child Development

    (1990)
  • R. Baker

    Land is life: A cultural geography of Australian contact history

    Australian Cultural Geographies

    (1999)
  • R. Barr et al.

    Developmental changes in imitation from television during infancy

    Child Development

    (1999)
  • S.R. Beck et al.

    Is tool-making knowledge robust over time and across problems?

    Frontiers in Psychology

    (2014)
  • S.R. Beck et al.

    Individual differences in children's innovative problem-solving are not predicted by divergent thinking or executive functions

    Philosophical Transactions of the Royal Society of London B: Biological Sciences

    (2016)
  • J. Belsky et al.

    From exploration to play: A cross-sectional study of infant free play behavior

    Developmental Psychology

    (1981)
  • D.F. Bjorklund et al.

    Object play and tool use: Developmental and evolutionary perspectives

  • J. Bock et al.

    Subsistence ecology and play among the Okavango Delta peoples of Botswana

    Human Nature

    (2004)
  • J. Bock

    Farming, foraging, and children’s play in the Okavango Delta, Botswana

  • R.M. Bongers et al.

    Geometries and dynamics of a rod determine how it is used for reaching

    Journal of Motor Behavior

    (2003)
  • K.S. Bourgeois et al.

    Infant manual exploration of objects, surfaces, and their interrelations

    Infancy

    (2005)
  • C.A. Caldwell et al.

    Identifying innovation in laboratory studies of cultural evolution: Rates of retention and measures of adaptation

    Philosophical Transaction of the Royal Society B

    (2016)
  • K. Carr et al.

    Eureka!: What is innovation, how does it develop, and who does it?

    Child Development

    (2016)
  • J. Chappell et al.

    The development of tool manufacture in humans: What helps young children make innovative tools?

    Philosophical Transactions of the Royal Society B-Biological Sciences

    (2013)
  • K. Connolly et al.

    The emergence of a tool-using skill in infancy

    Developmental Psychology

    (1989)
  • R.F.A. Cox et al.

    Action planning in young children's tool use

    Developmental Science

    (2006)
  • N. Cutting

    Children's tool making: From innovation to manufacture

    (2013)
  • E. Flynn

    Investigating children as cultural magnets: Do young children transmit redundant information along diffusion chains?

    Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences

    (2008)
  • N. Garon et al.

    Executive function in preschoolers: A review using an integrative framework

    Psychological Bulletin

    (2008)
  • S. Gaskins

    Children's daily activities in a Mayan village: A culturally grounded description

    Cross-Cultural Research

    (2000)
  • D.C. Geary

    The origin of mind: Evolution of brain, cognition, and general intelligence

    (2005)
  • D. Gentner et al.

    Structure mapping in analogy and similarity

    American Psychologist

    (1997)
  • Cited by (30)

    • Learning versus reasoning to use tools in children

      2021, Journal of Experimental Child Psychology
      Citation Excerpt :

      In this context, two main factors can explain some discrepancies concerning the age at which a child can solve a physical problem. The first is the child’s developmental trajectory, which is based on the physical events the child can experience over the years and which is mediated by the child’s own individual exploration as well as his or her own social and cultural environment (for studies exploring the effect of culture, see, e.g., Neldner et al., 2017, 2019). Thus, this interindividual variation can explain why some 5-years-olds performed better than chance in the Transparent–No Cue condition The second factor is the difficulty of the task in terms of physical understanding, which nevertheless remains necessarily determined by the first factor.

    • Preschool children's use of perceptual-motor knowledge and hierarchical representational skills for tool making

      2021, Acta Psychologica
      Citation Excerpt :

      We believe that this study, in accordance with other studies in the literature, will pave the way for more systematic investigations of the development of tool innovation and tool manufacture considering social and individual learning effects through the process of tool making in children. Even though a great interest in the topic arose in the last decade, results still were based on specific tasks, specific tools, and specific cultural groups (but see, Lew-Levy, Pope, Haun, Kline, & Broesch, 2021; Neldner et al., 2017, 2019). This shortcoming can only be resolved with multi-lab collaborations using different tasks, different tools, and various cultural groups.

    • Cognitive prerequisites for cumulative culture are context-dependent: Children's potential for ratcheting depends on cue longevity

      2021, Journal of Experimental Child Psychology
      Citation Excerpt :

      Furthermore, Osiurak et al. (2020) showed that a teacher’s theory-of-mind ability predicted cumulative performance, but only when the teacher did not have visible access to a learner’s actions. Work focussed on the constraints which may underly the tool innovation abilities of individual children, rather than the cumulative result of a social transmission process, is also relevant here; Neldner et al. (2017) provide evidence that the affordance visibility of a tool (whether one can easily perceive how the tool can be used to achieve a desired action e.g. presence of a visible hook for hooking) may affect the likelihood that task-relevant innovations to the tool occur. The examples outlined above thus raise the possibility that, for a population sharing a common pattern of cognitive abilities and behavioural proclivities, transmission patterns consistent with cumulative culture may be possible for some behaviours but not others.

    View all citing articles on Scopus
    View full text