Abstract
Recognition of future utility is a critical step in elevating a mere problem-solution into an innovation. It motivates us to engage in behaviors like keeping useful tools for repeated use, sharing them with others who need them, and refining their design—for instance—to maximize efficiency or durability. While previous research has assessed children’s capacity to use and create solutions to novel problems, we do not know when they become able to recognize the future utility of such solutions. Across three experiments, we explored the development of this capacity in 4- to 9-year-olds (N = 282) using vignette-based tasks. When asked to share a tool with another, children consistently selected the item with future utility above levels expected by chance alone. This behavior was found to increase with age in Experiments 2 and 3. Each experiment also included a series of secondary measures examining children’s understanding of other future-oriented behaviors—such as deliberate practice, deliberate study, mobile container use, and mental rehearsal. There were consistent positive correlations between performance on the tool-sharing vignette and children’s overall performance across the foresight measures, independent of age. This suggests that these are all expressions of children’s maturing foresight.
Keywords
Introduction
Children begin to use tools from an early age. Between 9 and 24 months, they learn to grip and use simple tools, like spoons and toy hammers, with increasing precision (Fragaszy et al., 2016; Keen et al., 2014; McCarty et al., 1999, 2001). Toddlers even demonstrate a degree of flexibility in using such tools to solve immediate problems, such as when 18-month-olds use a rake tool to retrieve out-of-reach objects (Brown, 1990; Rat-Fischer et al., 2012). Around this age, children also develop competencies in object construction and can stack, nest, and affix items to create increasingly complex structures (Marcinowski et al., 2019). However, children’s capacity to use familiar objects to solve novel problems appears to develop much later. For instance, it is only by age 8 that children tend to independently retrieve a floating object from a tube by adding water or other small items to raise it within reach (Cheke et al., 2012; Hanus et al., 2011; Nielsen, 2013). Similarly, only 8-year-olds and older children typically show a capacity to modify objects or materials to construct new tools, such as devising a hook tool from a straight pipecleaner to lift a bucket from a tube (Beck et al., 2011; Cutting et al., 2011; Nielsen et al., 2014), although slightly younger children can solve such problems if the target materials have clearer affordances (Neldner et al., 2017), are more familiar (Gönül et al., 2021), or are made more salient (Ebel et al., 2019; Miller et al., 2017), and when given additional time (Voigt et al., 2019) or the physics of the task are less complex (Breyel & Pauen, 2021).
This capacity to create novel solutions, by either devising a new tool or simply using something old in a new way, is often considered a core component of innovation in the developmental literature (Carr et al., 2016; Cutting et al., 2011; Neldner et al., 2019; Rawlings & Legare, 2021; Rawlings, 2022). Causal reasoning, executive functions, planning, creativity, and imagination (see Gönül et al., 2021; Rawlings & Legare, 2021) are just some of the many mechanisms that have been argued to play a role in children’s ability to generate solutions. However, innovations are not merely new tools or behaviors but are instead solutions that continue to be useful for us over time. Take the development of Post-it notes as an example. The adhesive used in Post-it notes was created not through the process of purposeful invention but was instead the undesired result of an experiment to create a strong adhesive. For years, the inventor of the adhesive—Spencer Silver—considered it to be a failure, until his colleague—Art Fry—needed to mark their place in a book and realized the weak, non-residue adhesive would make the perfect solution. Here, the innovation was not in Silver’s creation of the adhesive itself, nor in Fry’s personal one-off use of it as a bookmark, but rather in Fry’s recognition that it was a solution to a problem, and that it would be useful again in the future. Indeed, recognizing the future utility of a solution has been argued to be a critical feature of innovation (Suddendorf et al., 2018).
This recognition motivates us to keep useful solutions for the future, rather than toss them aside when the current problem is solved. It may also motivate us to refine the solution, so that it may be more durable for multiple future deployments or so that it is more efficient or easier to produce. Furthermore, once the future utility is recognized, one may also see the benefits it may offer others. Hence, one may want to share the solution with others in one’s community, whether it is for altruistic reasons or to gain financial or reputational rewards. For example, once the potential of the Post-it note adhesive was recognized, the company patented the temporary sticking agent; they then set about improving and diversifying the product itself (e.g., consumers are now spoilt for choice when it comes to their Post-it note color and size preference); and introduced it to local markets—and eventually the world—for monetary gain. Thus, recognizing the future utility of a solution is essential to such a transformation from a simple, one-off solution to an innovation creating technological and cultural change in populations.
From this perspective, then, emerging foresight capacities (i.e., the ability to consider future possibilities and adjust present behaviour accordingly; see Suddendorf & Moore, 2011) enable children to become innovative. When children recognize the future utility of solutions they may be driven to spend effort retaining, refining and sharing them for future use (Suddendorf et al., 2018; von Hippel & Suddendorf, 2018). And such behavioral expressions of innovation can, in turn, provide clues about development. Young children do seem to retain, at least in their memories, an understanding of how to solve a problem and an ability to use that solution at a later date. Beck and colleagues (2014), for instance, found that once 4- to 6-year-olds had observed the solution to the hook task, they could readily re-create that solution three months later. But, of course, memory of a solution alone does not imply that these children intentionally retained this knowledge in anticipation that they would reencounter the problem again in the future. Similarly, young children have also been recorded to maintain collections of items like shells, rocks, and toys (Burk, 1900; Lekies & Beery, 2013; McAlister et al., 2011), but it is unclear to what extent such hoarding is driven with future use in mind.
The development of children’s foresight more broadly, however, has recently attracted considerable research interest (e.g., Atance, 2018; Hudson et al., 2011; McCormack & Atance, 2011; Nyhout & Mahy, 2023; Suddendorf et al., 2022). One study, for instance, found that some 3-year-old children understand that they can improve their ability to retain objects by using mobile carrying devices (i.e., a basket; Suddendorf et al., 2020). Children of this age also begin to “budget” for the future, in that they will save resources for greater, future enjoyment rather than wasting such resources on a less rewarding, present activity (Atance et al., 2017; Jerome et al., 2023; Kamawar et al., 2019; Metcalf & Atance, 2011). By age 4, children also demonstrate some ability to correctly identify and obtain a solution for a one-off future problem (Redshaw & Suddendorf, 2013; Suddendorf et al., 2011), and by age 5 they can do so even in circumstances where both the target and distractor items are associated with past reward (Dickerson et al., 2018). Beyond physical solutions, 5- to 6-year-olds also show an emerging capacity to selectively acquire skills and knowledge that will be useful to them in the near future (Brinums et al., 2018, 2021; Casey & Redshaw, 2022; Davis et al., 2016). But it is unclear when in development children would also act to retain a solution (i.e., keeping possession of a physical tool or maintaining skills and knowledge via continued practice and revision) if they had reason to expect the problem would reoccur in the future.
To the best of our knowledge, no studies have examined children’s ability to refine an already functional tool to make it more efficient in solving a relevant future problem. Studies have, however, examined children’s basic ability to manipulate non-functional task materials to solve present problems. The traditional hook task, for example, tests children’s basic ability to take a straight pipecleaner and bend one of the ends to make a functional hook (Beck et al., 2011). Between ages 3 and 5, children show increasing competency when making tools by adding or subtracting extra materials to make it the correct length (Neldner et al., 2019; Voigt et al., 2019). In one recent study (Burdett & Ronfard, 2023), 4- to 12-year-old children were presented with a collection of materials and instructed to use them to retrieve items from a box. Results showed that children who iteratively modified the materials were more likely to retrieve the items than children who did not. Critically, however, even in this case the task could not be solved without manipulating the materials in some way, which differs from the form of refinement under consideration here: modifying an operational tool to increase overall future returns, for instance, by making it multifunctional, altering its design to minimize possible consequences that could arise from its use, or improving its durability, cost effectiveness or efficiency.
In contrast to tool refining, children’s basic inclination for sharing has been widely investigated (e.g., Brownell, 2013; Warneken, 2015). In one influential study, for instance, even 18-month-old infants showed a predilection toward handing an out-of-reach object to an adult who wanted to use it (Warneken & Tomasello, 2006). Other research has shown that 3-year-old children tend to share functional tools rather than non-functional tools when asked to help another agent perform a present-oriented task (Corriveau et al., 2017; Martin & Olson, 2013). Furthermore, there is some evidence to suggest that 5-year-old children (but not 3-year-olds) selectively share resources with a partner in anticipation that the partner will reciprocate in the future (Sebastián-Enesco & Warneken, 2015), and indeed such a tendency is associated with children’s performance on tasks measuring future-oriented thinking (Grueneisen et al., 2023). These studies are typically framed in the context of explaining the emergence of prosocial behavior in our extremely social species. Again, however, such tasks do not measure children’s ability to share a solution with the intention that the other person will receive a delayed benefit, which can facilitate the spread of innovations among a group or culture.
The Current Research
Across three studies, we investigated children’s capacity to recognize the future utility of a solution. To do this, children completed a picture-book vignette task assessing their understanding of related behavioral expressions which included retaining, sharing, or refining an item in anticipation of a future problem. We also included a series of shorter scenarios examining other future-oriented behaviors such as deliberate practice, information seeking, mental rehearsal, and the use of mobile containers (Brinums et al., 2018, 2021; Davis et al., 2016; Dunbar et al., 2005; Suddendorf et al., 2020), to create a measure of children’s foresight understanding. The aim of these studies was to explore the developmental trajectory of children’s understanding of future utility and its relation to other measures of foresight.
Experiment 1
Method
Participants
The final sample included 120 children between the ages of 4 and 7 years (M = 71.81 months, SD = 13.96 months, range = 49–95 months, 68 girls) recruited from the general public at a museum located in Brisbane, Australia. An additional 17 participants were tested but excluded from analyses due to shyness (n = 4), a diagnosed developmental disorder (n = 5), lack of attention (n = 5), or having watched a previous participant (n = 3). A post hoc power analysis indicated that the final sample of approximately 40 children per condition yielded a 91.6% chance of detecting large age effects (equal to r = .50). Fifteen children were missing data for all foresight vignettes but were included in the main picture-book task analyses (see Supplemental Materials). Ethics approval for the study was granted by the University of Queensland’s Health and Behavioral Sciences human research ethics sub-committee (Experiments 1 and 2 approval number: 2020/HE001485; Experiment 3 approval number: 2021/HE000680).
Materials and Procedure
Picture-Book Task
To examine children’s recognition of future utility, they completed a picture-book vignette on an iPad which followed a character who could select a toy to solve a future problem. The six toy options included a wagon (target), ball, baton, scoop, kite, and a book, yielding a chance value of 16.7%. To ensure children were familiar with the toys, they watched a short video prior to the picture-book task which labeled and described a potential use for each toy (see Figure 1 and Supplemental Materials). The study was a between subjects design with children receiving one of three variants (counterbalanced by age in years). Children either received a story where the character could retain a toy for the future (n = 40, Mage = 72.10 months, SDage = 14.27 months); share that toy with another child who could use the toy later (n = 38; Mage = 71.48 months, SDage = 13.54 months), or the character could modify the tool to improve it for future use (n = 42; Mage = 71.87 months, SDage = 14.45 months).

Stimuli Presented in One of the Three Toy Description Variations (Top Left), the Retain Condition Item Selection (Top Right), the Share Condition Item Selection (Bottom Left), and the Refine Condition Item Selection (Bottom Right). The three item-selection panels (top right, bottom left, and bottom right) were in-story depictions of the possible choices. Children made their selection using an interactive interface where only the items were displayed on a white background (order randomized).
All three stories begun by introducing a character, Max, who was explained to enjoy collecting shells at the beach but could not bring many home as he could only carry them using his hands. To establish children’s baseline capacity to recognize the solution to a current problem, children were then asked to select which of Max’s six toys should be taken on his next trip to the beach and to explain why they chose that item. Regardless of children’s selection, the story continued with Max taking the wagon (target item) to the beach and using it successfully to carry more shells. From this point in the narrative onwards, the three stories differed depending on condition, but all followed a similar structure.
For children assigned to the retain condition, Max then returned home and spent time reading his book. Next, it was explained that Max and his family would be going to the beach again the following weekend and, as the car was nearly full, Max could bring only one of his six toys to the beach (see Figure 1). Children were asked to select which toy he should keep in the car and explain their reasoning.
Conversely, children in the refine condition were told that Max returned home from the beach and brought his wagon full of shells up to his room. But, because the shells were full of seawater from the ocean, the seawater pooled up in the wagon and then made a mess of Max’s room, making him upset. The narrative then continued with Max reading his book. It was then explained that Max and his family would be going to the beach again the following weekend, but before they went, Max could make alterations to his wagon. The options included making a small hole in the wagon’s base (which was the target as it would allow the water to drain away), making the handle longer, cutting off the handle, cutting the wagon in half, painting the wagon red, or leaving the wagon unchanged (see Figure 1). Children were asked which of the six options Max should choose and the reasoning behind their selection.
Finally, after making their initial toy selection, children in the share condition were introduced to Max’s friend Molly who also liked to collect shells but did not have a large shell collection as she could not hold many in her hands to carry home. Max and Molly then read a book together before it was explained that Max needed to help Molly pack by telling her which of her six toys she should take (see Figure 1). Children were asked to choose which of the six toys Max should tell Molly to pack, and why they selected that toy.
For all picture-book item selections, children’s responses were coded as passing only if they selected the target item. If children justified their choice by mentioning the utility of the item within the relevant problem’s context, then their verbal responses were coded as passing. Potential passing responses for the baseline, retain, and share item selections included “to collect more shells” or “it can carry lots of stuff.” Responses such as “to let the water drain out” or “so Max won’t make a mess” were scored as passes for the refine selection.
Foresight Measures
To assess the development of related future-oriented behaviors, a series of scenarios and corresponding questions were constructed based on previous research (Brinums et al., 2018, 2021; Davis et al., 2016; Suddendorf et al., 2020). Two distinct scenarios, which were either an open-response or forced-response format, were presented for each behavior totaling eight scenarios overall (see Supplemental Table S1). The open-response format asked children to imagine a future scenario and describe what they would do in this context. The forced-response format introduced two characters, one who demonstrates the respective capacity (i.e., practice, study, container use, or mental rehearsal) and the other who does not, and asked children to identify the character mostly likely to succeed on a relevant task. The four open-response scenarios (order randomized) were presented first, followed by the four forced-response scenarios (order randomized).
All answers to the forced-response measures were coded as passing if children selected the target character. For the open-response questions, responses describing the relevant target behavior were coded as passing. For example, responses such as “practice lots” and “do training” on the deliberate practice question, “ask someone” or “learn it” on the deliberate study question, “bring a bag” on the mobile container question, and “think about it over and over” on the mental rehearsal question would be considered as passing. However, for the mental rehearsal question, responses referencing other useful future-oriented strategies such as cognitive offloading (see Risko & Gilbert, 2016) were also coded as passing (e.g., “write it down” or “say it over and over”). Blind inter-rater reliability coding on approximately 20% of the sample (24 participants) indicated a percentage agreement of 91% across items. An overall foresight score was created by calculating the proportion of correct responses provided by children.
Results and Discussion
To test whether there was an effect of gender on children’s performance across the picture-book selection items, independent of age, we performed a series of logistic regressions. All comparisons but one (see Table S3 in Supplemental Materials) were not significant, and the single significant effect did not change the statistical relationship between age and the variable of interest. Data were therefore collapsed across gender for all analyses reported below. Children’s performance on the three picture-book tasks is described in Tables 2 and 3.
Overall, children were significantly above chance in identifying the wagon as the item Max should take to the beach when there was no delay between the introduction of the problem and the item selection (see Table 1). Performance was associated with age, with older children selecting the wagon more often than younger children and, as expected, all age groups performed above chance (see Table 2). Nonetheless, even on this basic assessment of children’s ability to recognize present utility, performance did not approach ceiling for any age group. For the retaining, sharing, and refining item selections, only children in the share condition selected the item with future utility above chance expectations (see Table 1). However, age was not associated with performance on the share item selection, and all age groups were above chance in identifying the wagon as the correct toy to share with Max’s friend (see Table 2). Age also did not correlate with performance on either the retain or refine item selection (see Table 1). Furthermore, children’s performance on the retain, share, and refine selection was not related to their earlier baseline performance. This suggests that children were not simply picking the wagon in the retain, refine, and share selections because they had chosen it earlier.
Children’s Picture-Book Performance and Inter-Task Correlations in Experiment 1.
p < .05, **p < .01, ***p < .001.
Number of Children Choosing the Target Item for Each Picture-Book Selection Opportunity in Experiment 1, by Age Group.
p < .05, **p < .01, ***p < .001.
Performance on each forced-response foresight measure was overall quite high, with children selecting the target character above chance expectations (50%) in all four scenarios (all ps < .001). There were also positive correlations between age and performance for all eight individual measures (see Table 3) aligning with the age-related improvements previously observed when examining these behaviors in children (Brinums et al., 2018, 2021; Davis et al., 2016; Dunbar et al., 2005; Suddendorf et al., 2020).
Children’s Performance on the Individual Foresight Measures in Experiment 1 and Their Overall Score Across All Foresight Measures.
Note. Individual foresight measures were scored as pass/fail where fail = 0 and pass = 1; the possible range for overall score was also 0 to 1, with no questions correct = 0, and all questions correct = 1.
Spearman’s Rho.
Point-biserial correlation.
p < .05, **p < .01, ***p < .001.
We also observed significant, positive correlations for 24 of the 28 possible comparisons between the individual foresight measures (see Table 4). Thus, as planned, we created an overall foresight score by combining children’s performance across the eight items. As expected from the age-related trends observed for the individual measures, performance on this combined score was also positively associated with age such that older children had higher overall scores than younger children (see Table 3).
Phi (Above the Line) and Pearson’s Age-Partialled (Below the Line & italicised) Correlations Between Individual Foresight Measures in Experiment 1.
p < .05, **p < .01, ***p < .001.
Children’s overall performance on the foresight measures assessing these future-oriented behaviors was not associated with their performance in either the retain or refine condition of the picture-book task (see Table 1). Overall performance was, however, positively associated with performance in the share condition independent of age-related effects (see also Table S4 in Supplemental Materials for correlations between share task performance and individual foresight items). This finding suggests that there may indeed be a common capacity underlying children’s performance across these measures: foresight.
Experiment 2
Although we found evidence of a relationship between children’s foresight and the sharing task, the lack of such associations for the retain and refine tasks may reflect some issues with how we attempted to measure these constructs. First, in the retain condition, it may not have been apparent that Max still wanted to collect more shells. Thus, we adjusted this story in Experiment 2 by including a sequence between Max’s first use of the wagon and the retain selection to make it clear to children that Max wants to collect more shells. Second, in the refine condition, children may have been reluctant to make changes that they perceived as damaging to the wagon. We considered various possible solutions to this issue (e.g., Max could make a plug for the drilled hole); however, each of these themselves presented new issues and, with the consideration of project time constraints, we did not include the refine condition in Experiment 2.
Furthermore, we identified a few potential wording issues with the foresight measures which may have impacted children’s interpretations of the questions and thus their performance. First, it is possible that children performed above chance on the forced-response measures without reasoning about the future by, for instance, relying on a simpler heuristic: more activity is better. In Experiment 2, we therefore included a non-target action in each scenario which the distractor character engaged in more often than the target character (see Supplemental Table S2). Second, many children described cognitive offloading (see Armitage et al., 2020; Risko & Gilbert, 2016) instead of mental rehearsal as the appropriate action to undertake in the mental rehearsal open-response measure. Given this, and the exclusion of this behavior’s forced-response counterpart (see Supplemental Materials), we also excluded the open-response measure of mental rehearsal in Experiment 2. Third, the context for the mobile container open-response measure was not clearly oriented in the future as children often described present-oriented actions in their responses (e.g., asking friends or parents to help them carry items in or taking the items one-by-one), thus a new context was created for Experiment 2 (see Supplemental Table S2). And last, while the deliberate study open-response measure gauged whether children understood the benefit in seeking out information, it did not allow them to demonstrate their understanding of the utility of repetitive study. As such, we devised a new scenario for the deliberate study open-response measure where the context required rereading of information as the solution (see Supplemental Table S2).
Method
Participants
The final sample included 80 children between the ages of 4 and 7 years (M = 71.94 months, SD = 14.68 months, range = 49–94 months, 43 girls) recruited from the general public at the same museum as for the previous experiment. An additional 14 participants were tested but excluded from analyses due to a diagnosed developmental disorder (n = 6), having English as a second language (n = 3), lack of attention (n = 4), or parental prompting (n = 1). As in Experiment 1, a post hoc power analysis indicated that the final sample of approximately 40 children per condition yielded a 91.6% chance of detecting large age effects (equal to r = .50).
Procedure
The overall procedure was identical to Experiment 1; however, as noted earlier, we excluded the refine condition, and thus, children were allocated (age roughly counterbalanced) to either the retain (n = 41; Mage = 71.49 months, SDage = 14.42 months) or the share (n = 39; Mage = 72.41 months, SDage = 15.11 months) condition. Furthermore, the story in the retain condition included an additional narrative sequence which was delivered immediately following Max’s first selection and use of the wagon. In this sequence, Max returned home from the beach and placed his newly collected shells on top of his dresser, but it was explained that he only had enough shells to cover one half of his dresser and needed more to complete his collection. In addition, children only received six foresight measures (rather than the eight in Experiment 1) targeting their understanding of deliberate practice, deliberate study, and mobile container use (see Supplemental Table S2 for updated wording). Blind inter-rater reliability coding for the open-response foresight items on approximately 20% of the sample (18 participants) indicated a percentage agreement of 94% across items.
Results and Discussion
There was no significant effect of gender; thus, data were collapsed across this variable for all following analyses (see Supplemental Table S5). Children again performed above chance in identifying the wagon as the solution to Max’s problem and performance also, again, improved with age (see Table 5). However, only children aged 5 and older selected the wagon above chance (see Table 6). As our sample size in this experiment was smaller than that in Experiment 1, this may reflect an issue of reduced power. In the share condition, children again selected the wagon as the toy Max should share with his friend above chance expectations, and performance on this task was positively associated with age (see Table 5). But, in line with their baseline performance, the 4-year-olds did not perform above chance on the share selection (see Table 6), and thus, this sub-sample may not have been representative of 4-year-olds more generally.
Children’s Picture-Book Performance and Inter-Task Correlations in Experiments 2 and 3.
p < .05, **p < .01, ***p < .001.
Number of Children Choosing the Target Item for Each Picture-Book Selection Opportunity in Experiments 2 and 3, by Age Group.
p < .05, **p < .01, ***p < .001.
Furthermore, it is notable that despite the attempt to increase the salience of Max’s desire to collect more shells, children again did not choose the wagon in the retain selection above chance expectations (see Table 5). While it is possible that the capacity to retain a solution for future needs is late to develop, it is more plausible that this measure underestimates children’s ability to retain a solution for repeated use and, as a consequence, their capacity to recognize future utility. Indeed, as reviewed in the introduction, even 3-year-olds have been shown to save resources for future deployment (Atance et al., 2017; Jerome et al., 2023; Kamawar et al., 2019; Metcalf & Atance, 2011) suggesting that they can in fact show this capacity in some contexts.
Descriptive data suggested that our changes to the forced-response and open-response vignettes largely worked as intended (see Supplemental Material). As in Experiment 1, children selected the target character in all forced-response measures above chance expectations (50%; all ps < .001). We observed positive correlations between age and performance for four of the six individual foresight measures (see Table 7). There were also significant, positive intercorrelations for 8 of 15 foresight comparisons (see Table 8). A number of these correlations between constructs were consistent across Experiments 1 and 2, typically involving the deliberate practice item, the mobile container forced-response item, and the open-response item assessing deliberate study.
Children’s Performance on the Individual Foresight Measures in Experiments 2 and 3.
Note. Individual foresight measures were scored as pass/fail where fail = 0 and pass = 1; the possible range for overall score was also 0 to 1, with no questions correct = 0, and all questions correct = 1.
Spearman’s Rho.
Point-biserial correlation.
p < .05, **p < .01, ***p < .001.
Phi (Above the Line) and Pearson’s Age-Partialled (Below the Line & italicised) Correlations Between Individual Foresight Measures in Experiment 2.
p < .05, **p < .01, ***p < .001.
Children’s overall foresight score was again positively associated with age (see Table 7). As in Experiment 1, we did not find a significant relationship between overall foresight scores and performance in the retain condition, though given the continued methodological issues with the retain condition as discussed above this may not be surprising. Promisingly, however, overall foresight was positively associated with their performance in the share condition, independent of age (see Table 5; see also Table S4 in Supplemental Materials for correlations between share task performance and individual foresight items). This replicates the main positive finding from Experiments 1 and so provides further support that these measures may indeed be tapping into children’s development of foresight more generally.
Experiment 3
We conducted a third experiment assessing children’s performance on the share task to see if we could replicate this result in a sample with a broader age range.
Method
Participants
The final sample included 82 children between the ages of 4 and 9 (M = 83.72 months, SD = 21.31 months, range = 48–119 months, 43 girls) recruited from our university’s database of families interested in participating in developmental research. An additional six participants were tested but excluded from analyses due to low engagement in the tasks (n = 3), parental prompting (n = 1), or diagnosed developmental disorders (n = 2). An a priori power analysis indicated that a sample size of 82 children would yield a greater than 80% chance of detecting medium effects (equal to r = .30). The experiment was conducted as part of a larger project published elsewhere (see Ockerby et al., 2025).
Procedure
Children completed the share picture-book task, identical to the one used in both Experiments 1 and 2. The series of six short foresight measures used in Experiment 2 were also again administered here (see Supplemental Table S2). For the open-response foresight measures, blind inter-rater reliability coding on approximately 20% of the sample (18 participants) indicated a percentage agreement of 98% across items.
Results and Discussion
There was no significant effect of gender; thus, data were collapsed across this variable for all following analyses (Supplemental Table S6). We again replicated that children performed above chance when identifying the solution to an immediate problem and that this was associated with age (see Table 6). Likewise, children also performed above chance on the future-oriented share item-selection task (both overall, and as individual age groups; see Table 5), with older children selecting to share the wagon at higher rates than younger children (see Table 4).
However, unlike previous experiments, children’s baseline performance was negatively correlated with performance on the share selection, independent of age. It is possible that children who did not correctly identify the wagon as the solution to Max’s problem may have benefited from seeing Max use the wagon to collect shells and, as a result, were better able to identify the wagon’s utility in the sharing context. Indeed, while 39% and 75% of respective 5- and 6-year-olds chose the wagon at the baseline item selection juncture, 100% of children in these age groups later chose to share the wagon. Conversely, children who selected the wagon at the first (baseline) opportunity may have been motivated to choose an alternative toy when given the second selection opportunity regardless of the selection context. And, in fact, looking at 8- and 9-year-olds performance across the two selection opportunities, 100% of children selected the wagon at the baseline juncture, but this dropped to 86% across the two age groups at the share juncture.
Children again performed above chance across all three forced-response foresight measures (all ps < .001), and performance on these tasks was positively associated with age (Table 7). Likewise, age and performance across the open-response foresight measures were also positively correlated. Furthermore, 6 of the 15 correlations between each individual foresight item were significant (Table 9). The consistently observed relationships tended to involve items assessing deliberate study and deliberate practice. This suggests that children’s understanding of deliberate study and practice as useful behaviors in preparing for the future could represent a more flexible capacity to engage in repetitive, goal-directed action, regardless of whether the action in question involves the building of knowledge or skills. Thus, children’s capacity to engage in these behaviors may emerge in tandem.
Phi (Above the Line) and Pearson’s Age-Partialled (Below the Line & italicised) Correlations Between Individual Foresight Measures in Experiment 3.
p < .05, **p < .01, ***p < .001.
Children’s overall foresight performance was positively associated with age (see Table 7), aligning with our results in both Experiments 1 and 2. Overall foresight performance and share task performance were positively correlated independent of age, replicating the results from Experiment 1 and Experiment 2 (see Table 4; see also Table S4 in Supplemental Materials for correlations between share task performance and individual foresight items).
General Discussion
The ontogeny of tool innovation has historically been examined by studying children’s ability to manufacture new solutions when faced with a novel problem (Beck et al., 2011; Breyel & Pauen, 2021; Burdett & Ronfard, 2023; Cutting et al., 2011; Neldner et al., 2017; Nielsen et al., 2014; Rawlings, 2022; Voigt et al., 2019). However, not all innovations require the construction of a new solution, and a newly devised tool can come to nothing if, after one use, it is simply thrown away. Accordingly, a critical aspect of innovation is the recognition that a solution has utility in the future (Suddendorf et al., 2018). This recognition of future utility is what motivates us to retain tools to use again in anticipation that a problem will reoccur, refine them to maximize their utility, and share them with others—thus facilitating their spread in a group or culture. Yet, little is known about the development of these behaviors in children. The aim of this paper was to investigate children’s capacity to recognize future utility through their understanding of future-directed tool retaining, sharing, and refining behaviors.
We encountered complications with both the refining and retaining vignettes we had constructed. Children performed quite poorly on these tasks and, in contrast to our hypothesis, performance did not improve with age. This does not, however, mean that children of the sampled age range are not able to recognize the utility of retaining or refining tools in preparation for a future problem, or that such recognition does not improve throughout early and middle childhood. Instead, it is more likely that the tasks underestimated these capacities in children, given the issues we had with operationalization. For instance, our attempt to assess children’s capacity for refining a tool involved drilling holes in a wagon. In hindsight, perhaps it was unsurprising that children were reluctant to pick this option. Future studies could try to operationalize refining in ways that do not run counter to norms against breaking a functional tool to achieve one’s ends or make the required modification reversable. Alternatively, it is also possible that children struggled in the refining task due to difficulties with recognizing a new utility of a familiar object to solve a novel problem (i.e., functional fixedness). While there is some evidence to suggest that familiarity with task materials can improve children’s creative problem-solving (Gönül et al., 2021), prior use of the materials for a different function has been shown to impede 6- and 7-year-olds’ performance on similar tasks (e.g., German & Defeyter, 2000). To better understand the impact of functional fixedness in tool refining contexts, future research could also vary the familiarity of the target object and its contextual function. Furthermore, to limit the impact of narrative repetition in tasks where children must identify that the same tool will be most useful across multiple occasions, it may be beneficial to modify the story so that all tool options have been useful in the past, but only the target tool should be retained to solve a specific future problem.
Despite these issues with our retaining and refining tasks, our experiments yielded a coherent set of results about children’s future-directed sharing. Across our three experiments, we found consistent age-related performance trends, providing compelling evidence that children’s capacity to recognize when a solution will be useful for others improves throughout early and middle childhood. There was mixed evidence for the age at which children can reliably identify an item with future utility, with 4-year-olds performing above chance in Experiments 1 and 3, but not in Experiment 2. This complements previous research which found critical transformations in other future-directed capacities between ages 4 and 5 (for reviews, see Atance, 2018; Hudson et al., 2011; McCormack & Atance, 2011; Nyhout & Mahy, 2023; Suddendorf et al., 2022), as well as the finding that children begin to selectively share when there is opportunity for future reciprocity around age 5 (Sebastián-Enesco & Warneken, 2015). Furthermore, while it is possible that the methodological changes to the foresight vignette tasks between Experiments 1 and 2 influenced results, we note that consistent age-related improvements across the individual items were observed, aligning with previous research studying their development (Brinums et al., 2018, 2021; Davis et al., 2016; Dunbar et al., 2005; Suddendorf et al., 2020). Performance across some of these measures were also inter-related, particularly for the items gauging children’s understanding of deliberate study and practice in Experiments 2 and 3, providing evidence that these future-oriented behaviors may indeed reflect children’s emerging capacity for foresight more generally.
Children’s performance on the sharing task was consistently related to their overall foresight score, independent of age, suggesting that performance on these measures is underpinned by a common mechanism: foresight. Foresight is purported to be central to the development of modern human culture (Suddendorf et al., 2022). Our capacity to mentally transport ourselves in time allows us to consider problems that we are likely to reencounter or even face for the first time. In doing so, we can then prepare for such problems by not only equipping ourselves with physical tools and specific knowledge or skills, but also instructing others to do the same. Indeed, Vale and colleagues (2012) argue that humans’ capacity for mental time travel not only facilitates instances of cultural innovation, but also the transmission of such innovations via targeted, high fidelity information exchange such as teaching. As teaching can be conceived of as the act of sharing knowledge and skills, exploring the development of future-oriented teaching may therefore be another critical avenue for research exploring children’s recognition of future utility. However, while these consistent findings appear promising, they are only correlational and thus there could be other reasons for this link, including general verbal reasoning ability or task specific reasons such as semantic or contextual similarity across the share task and foresight vignettes. All tasks, for instance, required children to follow narratives involving third-party characters confronting future-directed problems, and it is possible that the narrative element, rather than the foresight element per se, was driving the correlations. Future research examining foresight and future-oriented sharing should thus utilize a variety of problem-solution scenarios including the use of behavioral tasks and independent measures of verbal reasoning ability to rule out such explanations.
In conclusion, these studies provide preliminary insight into understanding children’s capacity to recognize future utility. Our findings lay the groundwork for further investigations of this capacity and the behavioral expressions of future-directed sharing, retaining, and refining. The vignette tasks developed here can also be used to inform direct behavioral measures of these abilities to determine when children have the capacity to not only understand the behaviors of future-directed tool retaining, sharing, and refining but also engage in these innovative behaviors for themselves.
Supplemental Material
sj-docx-1-jbd-10.1177_01650254251344596 – Supplemental material for Young children’s understanding of future utility: A series of vignette tasks
Supplemental material, sj-docx-1-jbd-10.1177_01650254251344596 for Young children’s understanding of future utility: A series of vignette tasks by Zoe Ockerby, Jonathan Redshaw and Thomas Suddendorf in International Journal of Behavioral Development
Footnotes
Acknowledgements
The authors thank Queensland Museum for their support of this project throughout the data collection period and all participating children and their parents/guardians.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by an ARC Discovery Project grant (DP210101572) awarded to JR and TS.
Informed Consent
Children and their caregivers provided consent prior to participation.
Data Availability Statement
Data will be made available on the Open Science Framework upon publication.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
