Week 2 – Why do children lack flexibility when making tools? The role of social learning in innovation.

This early draft was authored by Nicola Cutting.

The capacity of humans to manufacture and use tools has evolved far beyond that seen in any other species (Boyd, Richerson & Henrich, 2011). This is thought to be due to our propensity for cumulative culture (Boyd & Richerson, 1996), where new techniques are copied throughout the social group and then improved upon in a ratchet-like effect (Tomasello, 1999; Tomasello et al., 1993). Cumulative culture involves two factors – termed ‘dual engines’ (Legare & Nielsen, 2015) – innovation and imitation. The literature has predominantly focused on how humans copy others – social learning, with children demonstrating faithful replication of techniques from a young age. Only in the last decade have researchers investigated novel inventions, or innovations. In contrast to human disposition for social learning, research has demonstrated that capacity for innovation is somewhat weaker. However, the majority of innovation studies test innovative ability at the individual level. The difficulty children demonstrate in these asocial problem-solving tasks may mask their ability for socially-mediated innovations (Rawlings & Legare, 2020).

This chapter will outline and discuss existing research into asocial tool-innovation in children in comparison to their ability to manufacture tools following social demonstrations. The main focus of the chapter will be on examining more recent studies of children’s socially-mediated innovations. Can children use socially-transmitted information flexibly to innovate novel tools?

How do tool techniques develop? The role of cumulative culture

[Definition and brief overview of cumulative culture if not covered in another chapter]

[Main theories suggest human uniqueness in cumulative evolution due to high fidelity social learning- active teaching and imitation.]

Whilst these lines of research have focused on the important questions surrounding the uniqueness of human cumulative culture and its evolutionary origins. The second component of cumulative culture – innovation has been somewhat neglected. Developmental Psychology is one approach that has been taken to investigate the potential origins of innovative ability in humans.

Children’s tool making: Independent (asocial) Tool Innovation

In contrast to children’s ability to faithfully replicate tool behaviours seen in others, children display great difficulty in innovating a simple tool for themselves. The most commonly used tool-innovation paradigm requires children to generate the idea of and manufacture a hooked tool from a pipecleaner in order to hook a bucket out of a tall narrow tube (Beck et al., 2011) (see Figure 1.). This task was based on a study conducted with New Caledonian crows (Weir et al., 2001), in which a female crow “Betty” spontaneously manufactured a wire hook to solve the task when her mate “Abel” flew away with the tool needed to solve the problem. Children aged between 3 and 10 were tested on the task, along with a ‘mature’ sample of 16-year-olds. When asked to retrieve the bucket from the tube, children had remarkable difficulty producing the required hooked tool to complete the task. Very few children under the age of 5 were successful, with success gradually increasing with age, with just over half of children successful at age 8 and 80% success at age 10 (see Figure 2.).

Children’s difficulty with the “hooks task” has been shown to be robust across a number of studies conducted by multiple research groups (Cutting, Apperly & Beck, 2011; Neldner, Mushin & Nielsen, 2017; Voigt, Pauen & Bechtel-Kuehne, 2009) and across cultures (Nielsen, Tomaselli, Mushin & Whiten, 2014). Similar levels of success were observed in a sample of African Bushman children whose culture necessitates a higher need to manufacture tools for themselves and consists of fewer pre-made tools than seen in Western society where the majority of tool-innovation research has been conducted.

The majority of studies investigating children’s tool-innovation ability have been based on the “hooks task” paradigm described above. However, a small number of other paradigms have been used and have generated similar findings. Cutting et al. (2011) found similar levels of success on a task requiring children to innovate a long straight tool needed to push a reward from a horizontal tube. Children’s difficulties innovating pipecleaner tools on the vertical tube (requiring a hook) and horizontal tube (requiring a long straight tool) tasks have also been shown to extend to studies requiring tools to be made from other materials (Cutting, 2013; Neldner et al., 2019; Voigt et al., 2009). The “Floating peanut” task requiring children to use water as a tool to retrieve a reward from a vertical tube by floating it to the top is a different paradigm used to test tool-innovation ability. This task has generated similar success levels to the vertical- and horizontal-tube tasks (Hanus, Mendes, Tennie & Call, 2011). Additionally, Mounoud (1996) found children aged four to nine had great difficulty constructing variously shaped tools from Lego in order to push a cube from one location to another inside a puzzle box. Together these studies suggest children’s asocial innovation difficulty to be a robust phenomenon.

Why is independent tool innovation so difficult?

To date, research has focused on how children’s tool-innovation ability may be constrained by cognitive capacity. It has been suggested that children’s poor performance is due to the ill-structured nature of tool innovation (Chappell et al., 2015; Cutting, Apperly, Chappell & Beck, 2014). Most problems we encounter in daily life are well-structured, they have clear start and goal states, and we simply choose between different options available to us. For example, in a tool-choice paradigm we have a start state of an apparatus containing a reward and two available tools, the goal state is to retrieve the reward, and the transformation to get from the start to goal state involves selecting the optimal tool to complete the task. In contrast, tool-innovation is ill-structured. The start and goal states are the same as in the well-structured example but there is little information of how to get from one to the other. The solver must generate and execute the solution for themselves (Jonassen, Beissner & Yacci, 1993; Reitman, 1965). To do this involves executive ability. One must inhibit actions that are incorrect, switch between different strategies and hold information about the problem in working memory. The difficulty of ill-structured problems is that they encompass all executive components in conjunction with each other and cannot simply be reduced to their component parts. The difficulty of ill-structured problems is demonstrated in studies with patients with frontal lobe damage and children with autism. These participants were shown to perform at typical levels for lab-based executive tasks tapping individual executive functions, but performed at comparatively lower levels in ill-structured tasks that required the use of multiple executive functions in conjunction with each other (Goel, Pullara & Grafman, 2001; Shallice & Burgess, 1991; White, Burgess & Hill, 2009). It is therefore likely that the protracted development of children’s executive abilities (Dumontheil, Burgess & Blakemore, 2008) may be a factor in their difficulty with ill-structured tool innovation tasks.

Learning to make tools from others: Imitation and Emulation

The above studies suggest that asocial innovation is very difficult for children. The design of these innovation studies also allows us to observe children’s capacity to manufacture tools following social learning. In these studies, if children were not successful at innovating the required tool for themselves then they next received a demonstration of how to manufacture the required tool, termed the ‘tool-creation demonstration’. This provided children with the opportunity to imitate the correct tool-making method. Beck et al. (2011) and Cutting et al. (2011) provided unsuccessful children with a demonstration in which the experimenter held her own pipecleaner horizontally and manipulated one end to form the required hook tool. Importantly in these demonstrations the experimenter did not show the correct orientation the tool needed to be in or enter it into the apparatus. Despite not being a full demonstration of how to complete the task, the vast majority of children quickly modified their own pipecleaner into the required hook tool and successfully retrieved the bucket from the tube.

These findings are in line with a wealth of research demonstrating that children easily learn how to manufacture their own tools by watching others. From around 30 months infants are able to manufacture a rattle toy consisting of three parts after watching a model (Barr & Wyss, 2008 ; Hayne, Herbert & Simcock, 2003; Herbert & Hayne, 2000).

Later tool-innovation studies included an additional demonstration phase for children. If children were unsuccessful at innovating a hook tool for themselves, they received what has been termed a ‘target-tool demonstration’. In this demonstration the experimenter showed children an example of the end-state tool, giving children opportunity to emulate making the tool needed for the task. As with the tool-creation demonstration described above the end-state target-tool was presented in a horizontal orientation and no demonstration as to how to use the tool on the task was offered. Target-tool demonstrations were included in a number of tool-innovation studies (Beck et al., 2014; Chappell, Cutting, Apperly & Beck, 2013; Cutting et al., 2014; Cutting, Apperly, Chappell & Beck, 2019) yielding modest improvements in children’s ability to succeed on the task. Children aged 6 to 7 were better able to emulate making a successful tool after seeing a target-tool example than younger children (Chappell et al., 2013). The majority of younger children required the full tool-creation demonstration in order to successfully complete the task.

Summary: Independent (asocial) innovation

So far, the presented research has shown that children find it extremely difficult to innovate simple novel tools for themselves and this is in contrast to their aptitude to learn how to manufacture tools from others. This fits with existing evidence suggesting the uniqueness of human culture is due to high fidelity social learning. But for culture (in this case tools) to evolve then innovation is crucial. Whilst innovations could occur by individuals in complete isolation, the difficulty of innovation and the structure of our social groups suggests this to be unlikely. So how do innovations occur?

Whilst tool innovation undoubtedly involves executive skills, and it is likely this that makes it difficult, we must consider whether the tool innovation paradigms discussed so far truly capture the nature of innovations that occur in real-life. These paradigms require children to work independently to create a novel solution they will not have encountered before. Although children will have experienced the properties of pipecleaners and have knowledge of hooks, the requirement to create this novel tool is very much a step-change. Whilst such step-change innovations do occur they are likely to be a rare form of innovation. Most innovations consist of smaller, progressive modifications that accumulate over time within a social environment. The next section explores research that has looked at the scaffolding of innovations and how children use social information to help them construct novel tools without the need for full demonstrations of tool manufacture.

Socially-mediated Tool Innovation: Can information from others help children to innovate tools?

Many studies have aimed to investigate the mechanisms underlying tool-innovation to try to establish where children’s difficulty lay. Although not the stated purpose of these studies, their design of providing information to children within a social context allows us to draw some conclusions surrounding socially-mediated innovations more akin to those likely to occur in real-life. This section will outline these studies and discuss the contribution they make to our understanding of children’s ability for socially-mediated innovations.

Affordances

Although children are presumed to have knowledge of the pliable properties of pipecleaners, this was confirmed by a study in which one group of children took part in a warm-up exercise manipulating pipecleaners by winding them around a pen and creating spiral shapes (Beck et al., 2011). Highlighting the affordance of the pipecleaners in this social manner did not improve innovation and was therefore taken as evidence that children already possessed knowledge of pipecleaner properties.

As stated previously, one of the main difficulties of the current tool-innovation paradigms is the high cognitive load placed on children. They must first generate the idea of a hook tool and then recognise the utility of the pipecleaner in allowing them to achieve this. The tool-choice paradigm presented by Beck and colleagues (2011) demonstrated that children could easily recognise the utility of a hooked tool, quickly and effectively using it to solve the task, however this task did not have an innovative component. Neldner et al. (2017) sought to reduce cognitive load whilst maintaining the need for innovation. In this study children were presented with a hook tool that had the non-hooked end curled over preventing it from entering the apparatus. The provision of the focal affordance (hook shape) as visual information reduces cognitive load as children were only required to recognise rather than generate the appropriate affordance of the material. Children aged 3 to 5 were nine times more likely to innovate a functional tool when the focal affordance was visible. However, successful innovation was still only seen in 45% of children (compared to 14% in the affordance non-visible condition), showing that although children were helped by the reduction in cognitive load and social information, innovation was still a difficult feat for young children.

Non-functional Tool Examples

Another attempt to scaffold children’s tool innovation was made in a study that presented children with the correct shaped but non-functional tool. Cutting et al. (2019) presented children with oversized pipecleaner hooks with which to solve the vertical-tube problem. However, rather than scaffolding innovation and acting as a prompt for creating the required tool, the presence of the oversized hook appeared to hinder children’s ability. In line with Beck et al. (2011) and Neldner et al. (2019) children easily recognised the affordance of the hook tool, choosing to use it significantly more than the straight pipecleaner. However, children were poor at modifying the non-functional hook into a functional tool, or manufacturing their own correctly sized hook from the straight pipecleaner provided. In fact, compared to a baseline condition where children received two straight pipecleaners, children who received an oversize hook and a straight pipecleaner were less likely to create a functional tool to solve the task. This therefore suggests that the presence of a correct but non-functional tool actually hindered children’s ability to solve the problem at hand.

A number of explanations for this finding were proposed. Building on Neldner et al.’s (2019) cognitive load theory, one possibility is that instead of acting as a clue to help children, the presence of the non-functional hook actually increased cognitive load. In contrast to the Neldner study where children simply needed to recognise the affordance of the hooked end of the material, in this study children needed to not only recognise the affordance of the hook but also realise that it was too big for the task and execute a plan of successful modification. These added requirements may have been more cognitively demanding than simply needing to recognise the solution and executing it for oneself.

Another possible explanation is that children’s behaviour was due to them being pre-programmed to learn from others, especially adults. Adults teach children and provide them with useful information, it therefore seems likely that children expect to receive useful information and help. They may therefore interpret the testing paradigm as a situation in which the adult present is likely to provide useful and relevant information and products. They may therefore expect that the materials they are given are ones that will be needed and will work to solve the task they are presented with. This disposition for social learning may hinder children in the context of innovation, because they are not expecting to innovate, they are expecting to be taught how to solve the problem rather than figuring it out for themselves.

Multiple scaffolds

Cutting et al. (2014) explored children’s ability to use information from others to innovate a hook tool. In this study half of children participated in a pipecleaner bending exercise prior to the innovation task to highlight the affordances of the materials. If unsuccessful on the tool-innovation task children then received a target-tool demonstration. The design of this study allowed researchers to assess children’s ability to use this social information regarding different aspects of the task. It was concluded that children’s main difficulty was with generating necessary information for themselves, i.e. that a hook is needed, that pipecleaners are pliable etc. When given this information children over 5 were able to coordinate it together to create a successful solution to the task. However, children under 5 lacked this flexibility and had great difficulty combining the different pieces of information even when presented by the researcher.

Summary: Socially-mediated innovation

The studies highlighted in this section demonstrate that although small improvements are seen in children’s ability to innovate in more socially supportive environments, children’s abilities still remain low. Young children demonstrate relatively inflexible behaviour in the domain of tool-innovation. Whilst tool-making following full demonstration is easy for children, children struggle to make innovations by modification. In some cases, attempts to support children’s innovation actually had the reverse effect, possibly due to children’s expectations for teaching from adults and reliance on imitation. [These ideas will be expanded on here].

Transfer of tool making knowledge

Asocial and socially-mediated innovation has been shown to be difficult for young children. The next question that arises is whether children show a lack of flexibility in all aspects of their tool behaviour.

Replicating a learnt technique

For techniques to prosper it is important that they are retained for future use (von Hippel & Suddendorf, 2018). The tool-rich world we live in today could not exist if we did not retain information learnt about how to make and use tools. At the first level it is important that once a new technique (e.g. making and using a hook tool) has been learnt, this technique can then be replicated for the same task on future occasions.

Children demonstrate excellent ability to manufacture identical tools on the exact same task following their own initial innovation. Whalley et al. (2017) presented children with three trials of the hooks task and found that children’s success on the task was stable across trials. Children who innovated a hook tool replicated their successful on subsequent trials. Whilst this ability to retain useful innovative information is reassuring, this task is limited in that the trials were presented in quick succession and we are only able to assess whether spontaneous innovations are retained for future use.

Beck et al., (2014) provide more substantial evidence for children’s ability to retain tool making knowledge. Children were tested on the hooks task twice with a three-month gap in-between. In the first presentation of the task children were recorded as successfully innovating a hook tool, or successfully manufacturing a hook tool following either the target-tool or tool-creation demonstration. Successful innovation was then measured at time two. Children retained knowledge of tool making over the three-month period, with the ability to manufacture the tool pre-demonstration in each session rising from 0 to 71% in 4- to 5- year olds and 16 to 68% in 6- to 7- year olds. There was no difference in success at time 2 depending on whether children spontaneously innovate at time 1 or received either of the demonstrations. However, low initial success rates and small sample size may be masking differences in how these factors affect retention rates.

Flexible use of learnt techniques

Whilst exact replication of a technique is important for the retention of that technique, to drive cumulative cultural evolution, techniques need to be transferred to new tasks. The distance between the original task and the new situation will determine how much the technique evolves. In close transfer tasks children demonstrate some ability to flexibly transfer knowledge of making hook tools to tasks that vary only in surface characteristics. Beck and colleagues (2014) presented children with three versions of the hooks task one after the other: the original clear tube and bucket, a shorter green tube containing a blue bucket with closed loop handle and a cuboid clear transparent box with a square yellow bucket. Each task was presented with its own different coloured pipecleaners and string distractors. On each version of the task children were given opportunity to innovate a tool for themselves and then received a tool-creation demonstration if necessary. Performance on the first task was low for all children, with older 5- to 6- year old children demonstrating better ability to flexibly transfer knowledge to the new tasks (5 to 86%) than younger 3 to 5 year old children (4 to 50%).

Children’s ability for far transfer was tested on task requiring them to retrieve rewards from the same apparatus using different materials. Children were unable to transfer their knowledge of hook making with one type of material (pipecleaners) to a second task using the same apparatus requiring them to create a hook tool using different materials (wooden dowels added together) and vice versa (Beck et al., 2014). Despite either independently solving the task by making a hooked tool or being shown how to make a tool, children were unable to use their knowledge of the tool required to make a successful tool from a new material.

Children’s lack of flexibility for far transfer is confirmed by studies requiring children to make two different tools on two different tasks. Knowledge of the affordances of the pipecleaner materials available did not help children to make their second tool after success was achieved (either independently or with social learning) on the first task (Cutting et al., 2011).

Together these studies demonstrate a lack of flexibility in children’s use of techniques once learned. Although children can retain their knowledge of a new technique over time, whether that knowledge was gained independently or learned from another person, children show great difficulty with using their knowledge flexibly to solve similar problems. Therefore, in addition to poor ability to independently innovate simple tools children also demonstrate great difficulty with making minor modifications to their learnt techniques.

General Summary

Children demonstrate a lack of flexibility in the domain of tools. Despite remarkable aptitude to learn how to make and use tools by imitating or being taught by others, ability to innovate simple tools has consistently been shown to be difficult. Initial studies into children’s tool innovation focused on independent, asocial innovation. These tasks yielded very low levels of success and were an important starting point in our understanding of children’s ability to innovate. However, it seems likely these studies do not give us true insight into how innovative abilities develop. Innovations are likely to be much more socially-mediated, and it is important that paradigms capture this. Some recent studies described in this chapter give us some indication as to how children use social information to make innovations by modification, and these suggest a complex picture. More work needs to be conducted in this area, along with new paradigms.

References

Barr, R., & Wyss, N. (2008). Re-enactment of televised content by 2-year-olds: Toddlers use language learned from television to solve a difficult imitation problem. Infant Behavior and Development, 31, 696-703.

Beck, S. R., Cutting, N., Apperly, I. A., Demery, Z., Iliffe, L., Rishi, S., & Chappell, J. (2014). Is tool-making knowledge robust over time and across problems? Frontiers in Psychology, 5. doi:ARTN 1395 10.3389/fpsyg.2014.01395.

Beck, S. R., Apperly, I. A., Chappell, J., Guthrie, C., & Cutting, N. (2011). Making tools isn’t child’s play. Cognition, 119, 301–306.

Boyd, R., & Richerson, P. J. (1996). Why Culture is Common but Cultural Evolution is Rare. Proceedings of the British Academy, 88, 73–93

Boyd, R., Richerson, P. J., & Henrich, J. (2011). The cultural niche: Why social learning is essential for human adaptation. Proceedings of the National Academy of Sciences, 108, 10918–10925.

Chappell, J., Cutting, N., Tecwyn, E. C., Apperly, I. A., Beck, S. R., & Thorpe, S. K. (2015). Minding the gap: A comparative approach to studying the development of innovation. In A. B. Kaufman & J. C. Kaufman (Eds.), Animal creativity and innovation (pp. 287–316). Academic Press.

Chappell, J., Cutting, N., Apperly, I. A., & Beck, S. R. (2013). The development of tool manufacture in humans: what helps young children make innovative tools? Philosophical Transactions of the Royal Society B: Biological Sciences, 368 (1630). Cutting et al., 2019    

Cutting, N. (2013) Children’s tool making: from innovation to manufacture (Doctoral Dissertation, University of Birmingham).

Cutting, N., Apperly, I.A., Chappell, J. & Beck, S.R. (2014). Why can’t children piece their knowledge together? The puzzling difficulty of tool innovation. Journal of Experimental Child Psychology, 125, 110-117.

Cutting, N., Apperly, I. A., Chappell, J., & Beck, S. R. (2019). Is tool modification more difficult than innovation? Cognitive Development, 52, 100811.

Cutting, N., Apperly, I. A., & Beck, S. R. (2011). Why do children lack the flexibility to innovate tools? Journal of Experimental Child Psychology, 109, 497–511.

Dumontheil, I., Burgess, P.W., & Blakemore, S.J. (2008). Development of rostral prefrontal cortex and cognitive and behavioral disorders. Developmental Medicine and Child Neurology, 50 (3), 168-181.

Goel, V., Pullara, D., & Grafman, J. (2001). A Computational Model of Frontal Lobe Dysfunction: Working Memory and the Tower of Hanoi. Cognitive Science, 25 (2), 287-313.

Hanus, D., Mendes, N., Tennie, C., & Call, J. (2011). Comparing the performances of apes (Gorilla gorilla, Pan troglodytes, Pongo pygmaeus) and human children (Homo sapiens) in the floating peanut task. PloS one, 6(6), e19555.

Hayne, H., Herbert, H., & Simcock, G. (2003). Imitation from television by 24- and 30-month-olds. Developmental Science, 6, 254-261.

Herbert, J., & Hayne, H. (2000). Memory retrieval by 18- to 30-month-olds: Age-related changes in representational flexibility. Developmental Psychology, 36(4), 473-484.

Jonassen, D. H., Beissner, K., & Yacci, M. (1993). Structural knowledge: Techniques for representing, conveying, and acquiring structural knowledge. Hillsdale, NJ: Lawrence Erlbaum Associates.

Legare, C. H. & Nielsen, M. (2015) Imitation and innovation: The dual engines of cultural learning. Trends in Cognitive Sciences, 19(11):688–99

Mounoud, P. (1996). A recursive transformation of central cognitive mechanisms: the shift from partial to whole representation. In Sameroff, A. J., & Haith, M. M. (Eds.). The five to seven year shift: the age of reason and responsibility, pp. 85–110, Chicago: Chicago University Press.

Neldner, K., Redshaw, J., Murphy, S., Tomaselli, K., Davis, J., Dixson, B., & Nielsen, M. (2019). Creation across culture: Children’s tool innovation is influenced by cultural and developmental factors. Developmental Psychology, 55(4), 877-889.

Neldner, K., Mushin, I., & Nielsen, M. (2017). Young children’s tool innovation across culture: Affordance visibility matters. Cognition, 168, 335–343.

Nielsen, M., Tomaselli, K., Mushin, I., & Whiten, A. (2014). Exploring tool innovation: A comparison of Western and Bushman children. Journal of Experimental Child Psychology, 126, 384–394.

Rawlings, B., & Legare, C. (2020). The social side of innovation. Behavioral and Brain Sciences, 43, E175. doi:10.1017/S0140525X20000217

Reitman, W. (1965). Cognition and thought. New York: Wiley

Shallice, T., & Burgess, P. W. (1991). Higher-order cognitive impairments and frontal lobe lesions in man. In Levin, H. S., Eisenberg, H. M., & Benton, A. L. (Eds.) Frontal lobe function and dysfunction (pp. 125–138). New York: Oxford University Press.

Tomasello, M., Kruger, A., & Ratner, H. (1993). Cultural learning. Behavioral and Brain Sciences, 16, 495–552.

Tomasello, M. (1999). The Cultural Origins of Human Cognition. Harvard University Press.

Voigt, B., Pauen, S., & Bechtel-Kuehne, S. (2019). Getting the mouse out of the box: Tool innovation in preschoolers. Journal of Experimental Child Psychology, 184, 65–81.

von Hippel, W., & Suddendorf, T. (2018). Did humans evolve to innovate with a social rather than technical orientation? New Ideas in Psychology, 51, 34–39.

Weir, A. A. S., Chappell, J., & Kacelnik, A. (2002). Shaping of hooks in new Caledonian crows. Science, 297, 981

Whalley, C. L., Cutting, N. & Beck, S. R. (2017). The effect of prior experience on children’s tool innovation. Journal of Experimental Child Psychology, 161, 81-94.

White, S. J., Burgess, P. W., & Hill, E. L. (2009). Impairments on “open-ended” executive function tests in autism. Autism Research, 2(3), 138-147.