Knowing How to Fold ‘em: Paper Folding across Early Childhood

Abstract

Folding paper is a seemingly simple act that requires planning, bimanual coordination, and manual strength and control to produce specific forces. Although paper folding has been used as an assessment tool and as a way to promote spatial skills, this study represents the first attempt to document when paper folding emerges across early childhood. Seventy-seven children (ages 18 months to 7 years) and an adult reference group (24 college-aged adults) completed three pre-specified folds on a single piece of paper. Dependent variables included whether children attempted each fold and, if so, the accuracy of each fold. Grip strength, pinch strength, and developmental level were examined as potential correlates of paper folding. The results demonstrated that paper folding emerges as early as 27 months of age but becomes more accurate with age. At least 50% of children between 4 and 5 1/2 years of age completed folds. Additionally, children with more age-appropriate problem-solving skills attempted more folds, independent of age. These findings provide a descriptive framework for the ages at which paper folding emerges and suggest that paper-folding interventions could be implemented at even earlier ages than what previously has been examined.

Introduction

The seemingly simple act of intentionally folding a piece of paper requires a complex integration of skills. A child must have fine-motor dexterity to pick up a piece of paper, steady it to make a bend, and reinforce the bend into a fold. This act requires the ability to plan and control a two-handed motor act in which each hand may be carrying out different motor plans to achieve the end goal. Similarly, intentionally folding a piece of paper requires the ability to spatially identify and plan the fold. If a series of folds are needed to make a more complicated design, such as a paper hat, even greater planning and spatial skills will be required. This idea is captured by Von Hofsten (2012) who argued that all actions are predictively controlled, meaning that goal-directed actions are performed based on knowledge of what is going to happen next. This knowledge may refer to the goal of the action or to the next action needed in a chain of sequential actions.

There has been substantial analyses of the emergence of many motor skills over the first few years of life, including detailed studies of tool use (Connally & Dagliesh, 1989; Connally & Elliot, 1972; Kahrs, Jung, & Lockman, 2013; Lockman, 2000; van der Kamp & Steenbergern, 1999), drawing (Braswell & Rosengren, 2008), and handwriting (Blote, Zielstra, & Zoetewey, 1987); however, we know of no studies that document changes in paper folding over the course of early development. While paper folding may not be viewed as an essential skill at first glance, it is a common task that may serve as foundational experience for later-developing visual-spatial skills that are associated with attainment and success in science, technology, engineering, and mathematics (STEM) disciplines (Humphreys, Lubinski, & Yao, 1993; Shea, Lubinski, & Benbow, 2001; Uttal et al., 2013; Wai, Lubinski, & Benbow, 2009; Wai, Lubinski, Benbow, & Steiger, 2010). Thus, describing when paper folding emerges in young children may help researchers to create and evaluate early interventions that promote visual-spatial skills and later STEM achievement. To this end, the purpose of the present study was to describe age-related changes in children’s paper folding from toddlerhood (18 months) to young school-age (7 years).

Paper Folding as an Assessment Tool

A number of tasks in childhood and adulthood require paper folding, from making arts and crafts to wrapping a present. Indeed, paper folding has been observed in the classrooms of U.S. elementary schools (McHale & Cermak, 1992), and paper folding has been used in subtests of standardized assessments to measure fine-motor (Bruininks & Bruininks, 2005) and visual-spatial skills (Youngstrom, Glutting, & Watkins, 2003). However, mental paper folding has a richer history in assessment than physical paper folding. In adults, mental paper folding tests have been used in armed forces qualifying exams since the 1950’s (Uhlaner & Bolanovich, 1952) and alongside other spatial tasks have been shown to relate to mathematical ability and to success in STEM fields (Humphreys et al., 1993; Shea et al., 2001; Wai et al., 2009, 2010).

At least one study (Embertson, 1987) has demonstrated a link between mental paper folding and training in physical paper-folding in adults. Specifically, undergraduate students completed a mental paper folding pre-test and then were randomly assigned to either 15 minutes of solving mental paper folding problems by folding a physical version of the model or by completing a control filler task. After the 15 minutes, participants completed a mental paper folding post-test. The results indicated that physically folding the paper models led to greater improvements in mental paper folding compared to the control task. These findings are consistent with the common-coding theory (Prinz, 1984) that posits that perception and motor actions may have a shared representation. As such, enhancements to the motor representation may facilitate the spatial representation of the folds. While only one study, the results suggest that physically folding paper into 3-D objects may benefit mental paper folding ability in adults.

Given the use of mental paper folding assessments in adults, a handful of studies have begun examining the ages at which young children are first able to perform mental folding tasks. In a study of children (4–7 years), the ability to perform mental paper folding tasks emerged around 5.5 years on average (Harris, Newcombe, & Hirsh-Pasek, 2013). While this was the average age across the group, there were substantial differences in the age at which individual children exhibited above-chance performance on this task. Moreover, mental paper-folding continues to improve across childhood (Ang & Lee, 2010), suggesting that mental paper folding skills emerge around the age of five years and continue to increase through elementary school. However, researchers know virtually nothing about when the physical act of paper folding develops in young children. Understanding age-related changes in physical paper folding is particularly important as physical paper folding is being used as an intervention to target cognitive and educational outcomes.

Paper Folding as an Intervention Tool

As an intervention tool, practice with paper folding has been associated with improvements in educational outcomes in school-aged children. For instance, paper folding was included as part of a sensorimotor intervention to improve handwriting in children (Oliver, 1990), and paper folding has been recommended for use in mathematics education to improve spatial abilities (Adams, 2000) and to facilitate multiplicative thinking (Empson & Turner, 2006). A number of researchers have also investigated the use of origami (i.e., a form of paper-folding art that originates from Japan) as an intervention to improve visual-spatial reasoning in school-aged children (Boakes, 2009; Cakmak, Isiksal, & Koc, 2014; Krisztián, Bernáth, Gombos, & Vereczkei, 2015; Taylor & Hutton, 2013). Some of these studies suggest that orgami practice enhances spatial skill in 4^th through 6^th graders (Cakmak, Isiksal, & Koc, 2014; Krisztián, Bernáth, Gombos, & Vereczkei, 2015; Taylor & Hutton, 2013). However, in a study of 7^th graders by Boakes (2009), origami-based training did not improve visual-spatial or geometry skills. The findings of this study raise the question of whether paper folding interventions may be more effective if implemented at earlier ages. Yet, very little is known about when basic paper folding skills emerge in children. This information is necessary in order to identify the age at which children can meaningfully undertake paper folding (either as an intervention or an assessment) as well as the different developmental skills that may serve as prerequisites for paper folding.

Developmental and Motor Correlates of Paper Folding

Early motor behavior serves as the foundation of many developmental achievements (e.g., Campos et al., 2000; von Hofston, 2012), and fine motor skills in children prior to school entry have been shown to be a strong predictor of later math, science, and reading performance (Grissmer, Grimm, Aiyer, Murrah, & Steele, 2010). Identifying the age-related changes of manual paper folding in young children, as well as potential contributers to performance of this behavior, may provide a framework through which to better understand and assess early fine motor skills that may pre-date later spatial abilities and academic achievement. One contributer to paper folding performance may be pinch and grip strength as these have been found to relate to handwriting in children (Alañiz, Galit, Necesito, & Rosario, 2015). Because paper folding and holding a writing utensil require similar manual strength, grip and pinch strength may also have a role in paper folding. While extensive norms have been collected for grip and pinch strength across development (Dodds et al., 2014), few studies in children have attempted to link these measures to functional skills. In older adults, declines in grip and pinch strength have been found to be related to a large range of functional skills (Taekema, Gussekloo, Maier, Westendorp, & de Craen, 2010).

At present there is minimal information regarding the ages at which physical paper folding appears in young children. This is a gap in our knowledge of the development of key motor and spatial abilities. The present study was designed as an initial attempt to describe age-related changes in paper folding in children from toddlerhood to early school age. Exploratory analyses examined paper folding as a correlate of basic motor skills (grip and pinch strength) and developmental level (Ages and Stages Questionnaire, 3^rd edition [ASQ-3]; Squires & Bricker, 2009).

Method

Design

In this descriptive design, paper folding was operationally defined in four ways: 1) the frequency of folds that children declined to attempt when asked, 2) the frequency of attempted folds that were made along the line, 3) the accuracy of the fold (i.e., distance of the fold from the intended line in centimeters [cm]), and 4) the frequency of different types of errors in folding (e.g., folding perpendicular to the line, folding the paper to the line and not along the line). To serve as a developmentally advanced comparison group, the descriptive statistics from the performance of the adult participants are reported but not included in the age models. Exploratory correlations examined paper folding in relation to motor measures (grip strength and pinch strength) and developmental level (ASQ-3).

Participants

This study was approved by the University of Wisconsin-Madison’s Institutional Review Board. Parents of participants under the age of legal consent signed a consent form before study activities began, and participants gave verbal assent at the time of participation. A total of 77 children (ages 18 months to 7 years, M = 3.92 years, SD = 1.33 years; 64.9% female) participated in the paper folding task, which was part of a larger study investigating children’s fine motor skills and drawing. In the larger study, children completed drawing tasks (shape drawing, person drawing, and free drawing) using different technology while also completing the present paper folding task and measures of grip and pinch strength. See Table 1 for additional participant information. Children were recruited from preschools and a children’s museum in a midsize Midwestern city. We had no control of the age of the participants who visited the museum, and we collected data from any child from 18 months to 7 years of age whose parents consented to the study. This resulted in a sample with few children under the age of 2 (n = 2) and over 6 (n = 5). We performed the analyses with and without the children in the outlying ages, and the same overall pattern of results was observed. Thus, age outliers do not account for the findings reported here.

Table 1.

Demographic information for all study participants, reporting the mean (M), standard deviation (SD), and range for all variables of interest.

Sample of Children
N = 77, 64.9% Female
	Mean (SD)	Range
Age (Years)	3.92(1.33)	1.83–7.70
Grip Strength (kg)	5.86(2.99)	0.0–12.0
Pinch Strength (kg)	2.55(1.11)	0.0–5.5
ASQ
Total	268.11(32.16)	140–300
Communication	56.67(5.11)	40–60
Gross Motor	53.22(8.06)	30–60
Fine Motor	51.00(9.27)	25–60
Problem Solving	57.33(5.16)	40–60
Personal Social	54.88(6.02)	35–60

Parents of the children were asked to complete an online demographic survey. Of the 48 (62.3%) parents who completed the survey, the majority (79.2%) were White/Non-Hispanic, and the mean years of education was 18.6 years (roughly equivalent to a Master’s degree). Parents reported an average of 7.4 (range: 4–10) for subjective social status using the MacArthur Scale of Subjective Social Status (Goodman et al., 2001). The MacArthur Scale asks respondents to place themselves on a ladder that represents people who have the least money, little or no education, and no job or a job that is not respected at the bottom (1) and people who have the most money, highest level of education, and highly respected jobs at the top (10).

In addition to the child sample, a comparison group of 24 college-aged adults (79.2% female) completed the task. The adults were recruited through personal contacts and signed a consent form at the start of the study.

Materials & Measures

Ages and Stages Questionnaire, 3^rd edition (ASQ-3) (Squires & Bricker, 2009).

The ASQ-3 is a 30-item, parent-report assessment tool that examines the developmental progress of a young child across five domains: communication, gross motor, fine motor, problem solving, and personal-social. We selected this instrument because of its ability to measure development as a function of age-specific milestones across multiple domains during early childhood. This would help inform us of whether a child may be performing below their same-aged peers in these key domains. With that said, we understand this measure’s limitations in detecting individual differences in development, as the ASQ-3 is most often used as a developmental screening tool. The ASQ-3 has different versions for children 2–60 months of age and takes 10–15 minutes for a parent to complete. In the present study we provided the parent the ASQ-3 version that matched the child’s chronological age. In all versions, parents respond to questions such as, “Does your child walk well and seldom fall?” by indicating “Yes” (10 points), “Sometimes” (5 points), or “Not Yet” (0 points). The scores are summed across each domain and then overall, with higher scores indicative of enhanced performance. Scores for individual domains range from 0–60; overall scores across the five domains range from 0–300. In reliability assessments (Squires & Bricker, 2009; Squires, Bricker, & Potter, 1997), the ASQ has demonstrated strong test-retest reliability (92% agreement across administrations), strong inter-observer reliability (93% agreement between parents and trained administrators), and acceptable internal consistency among items (Cronbach’s alpha = .51-.87; .60-.84 for the present age range). The ASQ-3 has also demonstrated high concurrent validity with other standardized tests.

Pinch and grip strength.

Pinch and grip strength were measured using a Preston Jamar hand dynamometer and pinch meter (Patterson Medical, Warrenville, IL). For grip strength, the smallest handle position was used for all participants. The researcher first showed the participant how the dynamometer worked, instructing the participant to stand with the elbow at a 90° angle with the elbow touching the side of the body while squeezing the dynamometer. The dynamometer’s weight was supported by the researcher’s hand. The researcher would say, “Squeeze as hard as you can for three seconds when I say go. Go, one, two, three,” in order to get the participants to grip for three seconds. Three measurements were attempted on each hand (six measurements total). Seventy-one children (92%) had at least one grip strength measurement. Fifty-nine children (77%) had all six grip strength measurements. The maximum of all the measurements (regardless of handedness) was recorded and used in the analyses. Pinch strength was measured using a lateral pinch grasp (i.e., key pinch with thumb on top). Participants were instructed on how to place their fingers on the pinch meter, and were asked to pinch as hard as possible. Three measurements were attempted on each hand, alternating left and right hands. Sixty-nine children (90%) had at least one pinch strength measurement. Fifty-eight children (75%) had all six pinch strength measurements. The maximum of all the measurements (regardless of handedness) was recorded and used in the analyses.

Procedure

Participants were tested individually in a reserved area (e.g., empty area of the classroom or dedicated research space at the museum) where they were instructed to sit in a chair across a table from the experimenter. While the experimenter interacted with the participant, an assistant operated a laptop camera to record participants’ hand movements during task completion. Upon request, parents were allowed to stay in the same room with their child but were asked not to interact with the child during the study.

The present paper folding task was designed to mimic aspects of other paper folding assessments (e.g., Bruininks & Bruininks, 2005) while additionally assessing folding in contexts that mirror more real-life folding tasks (i.e., folding consecutive folds in the same piece of paper) and adapting this task to make it feasible for young children through simplified instructions and demonstration of a fold. In this task, participants were given a single sheet of paper (21.59 cm x 13.97 cm) with four dotted lines drawn on the paper (see Figure 1). The experimenter demonstrated the folding task by tracing along the line (marked “Demo”) and folding one corner of the paper along the line, saying, “I am going to fold along this line like this.” The experimenter then turned the paper face up (i.e., the side with the lines on it was placed facing up), pointed along another corner line (Fold 1), and said, “Now, can you fold along this line?” If the child declined to fold in response to this question (i.e., by not picking up the paper or verbally indicating that it looked too difficult), the experimenter encouraged the child to attempt the fold. If the child did not fold the paper in response to this prompt, the experimenter moved on to the next fold or activity. After each fold was completed, the experimenter provided positive feedback. A similar procedure was used for both Fold 2 and Fold 3, but without any specific demonstration fold. Twelve children (16%) did not complete any of the folds, while five children (6%) completed only one of the three folds, eight children (10%) completed two of the three folds, and 52 children (68%) completed all three folds. After conclusion of the paper folding task, participants received a small gift for participation in the study.

Figure 1.

The paper folding sheet (top) and examples of scoring fold accuracy (bottom). The paper folding sheet included the demonstration fold by the researcher (“Demo”), Fold 1, Fold 2, and Fold 3 (completed in that order). Participants were instructed to fold along the dotted line. Fold distance was measured at the paper edge. When the fold was to the left of the intended line, the measurement was recorded as a negative. Then, absolute values of these measurements were used in the calculation of average fold distance.

Data Coding

Two independent coders watched video recordings of participants’ hands in order to determine the participants’ actions while completing the paper folding task. Coders first examined whether the child declined to attempt making a fold in the paper. Then, coders recorded information about how the participant manipulated the paper (folded it correctly, folded the paper edge to the dotted line, rolled the paper, etc.), the orientation of the paper (the to-be-folded line facing up or facing down), the location of the paper during the fold (on the table, in the air, or manipulated in the air without folding), and the location of any measurable fold that was ambiguous when looking at the paper data. Coders also determined whether a fold was measurable. A measurable fold was defined as an attempted fold along the line that made a crease from which the distance to the intended line could be measured. For folds that were deemed measureable, coders measured the distance of each measurable fold from the line using a ruler on the folded paper. Specifically, we measured the distance of the fold from each end of the line, and we averaged the two numbers (see Figure 1).

One additional coder coded 90% of the data using a simplified coding scheme, examining only refusal, fold measurability, and why the fold was not measurable. The coders demonstrated .98 inter-rater reliability. The few discrepancies in coding were corrected with consensus coding. When fold-accuracy measurement inconsistencies were within 0.1 cm of each other, the average of the discrepant measures was used. All codes were entered into a Qualtrics survey designed specifically for this experiment that prompted the coders to fill in answers for all of the previously mentioned categories. If the participant performed some action that was not pre-specified by the coding survey, the coder wrote down what occurred at the end of the survey. Because the adult data required less coding for paper folding errors, half of the videos for the adults were coded by Coder 1, and the other half were coded by Coder 2.

Data Analysis

All statistical analyses were performed in R version 3.2.2 (R Core Team, 2015). Multiple analyses were performed to examine whether children of different ages would demonstrate differing levels of paper folding abilities. We have treated chonological age as continuous variable rather than as a grouping variable in all analyses. Although it is common practice in developmental research to treat age as a grouping variable, methodologists (e.g., Cohen, 1983; Humphreys, 1978; MacCallum, Zhang, Preacher, & Rucker, 2002) have argued that this practice reduces statistical power and in some cases may lead to significant results that are not in fact significant (DeCoster, Iselin, & Galluci, 2009). As we had relatively few children at both age extremes (e.g., only two children under the age of two, and five over the age of six) we conducted our analyses twice, first with our entire sample, and then with these 7 children removed from the analysis. The results were equivalent, so we report the results that include the entire sample.

In the analyses, we first examined whether or not children declined to attempt the fold as a function of chronological age by conducting a binomial logistic regression that predicted refusal (declined or did not decline) based on age. We then examined whether or not the children were able to produce measurable folds (given that an attempt was made) as a function of chronological age by performing binomial logistic regression that predicted a measurable fold (measurable or not measurable) based on age. In our third set of analyses, we examined the frequency of the different errors that prevented a fold from being measurable using descriptive statistics. Lastly, we examined paper folding accuracy for folds that were deemed measurable by conducting a linear regression that examined the average distance of the fold from the line (in cm) as a function of chronological age. In all models, we examined whether the child’s sex (male or female) predicted folding ability. However, in all but one instance, sex was not a significant predictor and did not strengthen the statistical model fit. Therefore, sex was omitted as an independent variable in all analyses except the one instance in which it was a significant predictor (reported below in the results).

To explore if basic motor skills (grip and pinch strength) and developmental level (ASQ-3) were associated with paper folding performance, we calculated Pearson correlations. Using the ppcor package (Kim, 2015), we performed follow-up partial correlations controlling for chronological age.

Results

Age as a Predictor of Declining to Attempt Folds

Across the three folds, a subset of the children declined to attempt folds, and we examined the likelihood of declining to attempt a fold as a function of age. For Folds 1 and 2, 16.2% (n = 11) declined (12 participants declined both Folds 1 and 2, one participant declined Fold 1 but not Fold 2, and one participant declined Fold 2 but not Fold 1). For Fold 3, 26.5% (n =18) declined. Three separate binomial logistic regressions focusing on each fold revealed that age significantly predicted whether the child declined Fold 1, (b = −1.69, SE = 0.54, p = .002), Fold 2 (b = −1.74, SE = 0.56, p = .002), and Fold 3 (b = −2.30, SE = 0.55, p < .001). In each case younger children declined more than the older children. No adults declined to make any of the folds.

Descriptors of How Participants Folded

Videos of folding were coded to examine how the participants engaged in the folding task (see online Supplemental materials for an example). Specifically, we coded whether the fold was performed on the table or in the air. For the adults, the majority folded the paper on the table: 75% (n = 54) for Fold 1 and 79% (n = 19) for Folds 2 and 3. The majority of the children also folded the paper on the table: 92% (n = 59) of children for Fold 1, 94% (n = 60) for Fold 2, and 88% (n = 50) for Fold 3. In the adults, whether the individual folded in the air or on the table did not affect the accuracy of Fold 1, t(22) = 0.45, p = .66 , Fold 2, t(22) = 0.44, p = .67, nor Fold 3, t(22) = 0.12, p = .90. In the children, there were too few participants in the non-table folding group to perform Chi-Square Tests to determine if folding on the table was associated with making a measureable fold. However, binomial logistic regression analyses suggested that folding on the table versus in the air was not associated with chronological age in the children (p’s for all folds > .35).

Age as a Predictor of Producing a Measurable Fold

The youngest age at which a child was able to complete a measurable Fold 1 was 2.32 years (median age = 4.27). The youngest age at which a child was able to complete Fold 2 was 2.32 years (median age = 4.96). The youngest age at which a child was able to complete Fold 3 was 3.81 years (median age = 5.20). To examine whether a measurable fold was produced as a function of age, we conducted binomial logistic regressions across the three folds (see Figure 2).

Figure 2.

Logistic regression plots examining whether the child was able to make a measurable fold as a function of age for Fold 1 (solid line with circles), Fold 2 (small-dashed line with triangles), and Fold 3 (large-dashed line with crosses). On the y-axis, making a measurable fold was a binary variable (0 or 1), and the fitted line for each fold shows the predicted probability of making a measurable fold as a function of age. For example, there was a 50% predicted probability that a child would a measurable Fold 1 at the age of 3.0 years.

For Fold 1, 73.7% (n = 42 of the 57 children who attempted the fold) produced a fold that could be measured. Age significantly predicted whether or not the child produced a measurable Fold 1, b = 1.94, SE = 0.53, p < .001, with older children producing a measurable fold more often than younger children. Examining Fold 2, only 40.4% (n = 23 of the 57 who did not decline) produced a Fold 2 that could be measured. Older children produced a measurable Fold 2 significantly more often than younger children, b = 1.65, SE = 0.49, p < .001. For Fold 3, 30.0% (n = 15 of 50 who did not decline) produced a fold that could be measured. Older children were more likely to produce a measurable Fold 3 compared to younger children, b = 2.20, SE = 0.68, p = .001. All adult participants produced measurable folds.

Age as a Predictor of Error Type

Of the children who did not produce a measurable fold, the frequency of different types of errors can be seen in Figure 3. The most frequent error types involved folding the edge of the paper to the target line rather than folding on the line, imitating Fold 1 when attempting to complete Folds 2 and 3, and bending the paper rather than creasing it. Less frequent errors included scrunching or rolling up the paper, and folding the paper in half or perpendicular to the target fold. One child ripped the paper. To determine if certain error patterns were associated with age, we performed binomial regressions for the three most common error types. Committing the most common error (edge-to-line) across folds was not found to be associated with age, b = −0.19, SE = 0.25, p = .44. However, overimitating Fold 1 and making a bend in the paper (but not a crease) were both negatively associated with age, b = −0.66, SE = 0.28, p = .02, and b = −1.34, SE = 0.40, p < .001, respectively.

Figure 3.

Frequency of error patterns in children during folding. An edge-to-line error was coded when the participant folded the edge of the paper to the line instead of folding on the line. An imitate-fold-1 error was coded when the participant continued to fold the corners of the paper (like in Fold 1) instead of completing Folds 2 and 3. A bend-not-crease error was coded when the participant’s folding resulted in a bend but not a crease in the paper. Scrunch or roll was coded when the participant scrunched the paper or rolled it instead of folding. A perpendicular error was coded when the participant folded perpendicular to the line. A fold-in-half error occurred when the participant folded the paper in half rather than complete the instructed fold along the line. A rip was coded when the participant inadvertently tore the paper while trying to fold.

Age as a Predictor of Fold Accuracy

To examine if the precision of the fold was a function of the child’s age, we conducted linear regression models for each fold (see Figure 4). For Fold 1, the average fold was 0.93 cm from the intended line (SD = 0.52 cm). Age was found to significantly predict distance of fold from the intended line, such that with each year of age, participants became an average of 0.30 cm more accurate in their folding ability, b = −0.30, SE = 0.90, p = .003. For Fold 2, the average fold was 0.87 cm from the intended line (SD = 0.92 cm). For Fold 2, age was found to significantly predict distance of fold from the intended line, such that with each year of age, participants became an average of 0.86 cm more accurate in their folding ability, b = −0.86, SE = 0.16, p < .001. For Fold 3, the average fold was 1.66 cm from the intended line (SD = 1.17 cm). For Fold 3, age was found to significantly predict distance of fold from the intended line, such that with each year of age, participants became an average of 1.71 cm more accurate in their folding ability, b = −1.71, SE = 0.65, p = .02. The assumption of normality was challenged with this analysis, but a follow-up Spearman’s rank correlation confirmed a significant relation between age and Fold 3 average accuracy, ρ = −.56, p = .03.

Figure 4.

Linear regression plots examining the average distance of the fold from the intended line as a function of age.

As a comparison point, the average fold measurement for adults was 0.08 cm (SD = 0.10) from the intended line for Fold 1 (92% smaller error compared to participating children), 0.06 cm (SD = 0.05) for Fold 2 (92% smaller error compared to children), and 0.10 (SD = 0.06) for Fold 3 (93% smaller error compared to children).

Developmental and Motor Correlates of Folding Behavior and Accuracy

To explore possible relations among the paper folding variables, basic motor skills, and developmental levels, we performed correlations and then follow-up partial correlations controlling for age. As can be seen in Table 2, stronger grip and pinch strength were associated with better performance in numeous aspects of paper folding prior to but not after accounting for chronological age. Of the ASQ-3 subscales, less developed problem solving skills were associated with more frequent declining to attempt a fold, both before and after controlling for age.

Table 2.

Pearson R correlations (not controlling for age) among the folding, motor (grip and pinch strength), and developmental level (Ages & Stages Questionnaire [ASQ-3]) variables

	Grip Strength (kg)	Pinch Strength (kg)	ASQ -3 Communication	ASQ-3 Gross Motor	ASQ-3 Fine Motor	ASQ-3 Problem Solving	ASQ -3 Personal Social
# of Folds Declined	−.45***, n = 71	−.35**, n = 69	+.11, n = 45	+.30*, n = 45	−.16, n = 45	-.42**, n = 43	−.16, n = 43
# of Measureable Folds	+.51***, n = 63	+.47***, n = 63	−.20, n =39	−.15, n = 39	+.11, n = 39	−.08, n = 37	+.09, n =37
Average Fold Accuracy	−.33*, n = 56	−.17, n = 56	+.05, n = 35	+.22, n = 35	−.05, n = 35	−.05, n = 33	+.07, n = 33

^*p < .05

^**p < .01

^***p < .001.

The only correlation that remained significant after controlling for age is bolded.

Discussion

In the present study we aimed to describe age-related changes in paper folding ability in children ages 18 months to seven years. Specifically, we examined the ages at which the children were most likely to decline to attempt a fold, were able to create a fold as instructed, and were able to produce folds with greater accuracy. We also examined the patterns of folding errors in those who were unable to create the fold as instructed as well as potential developmental and motor correlates of paper folding ability. Overall, there were three main findings. First, our descriptive results suggest that children as young as two years old were able to make folds along a line, but folding ability was shown to improve with age with at least half of the children successfully completing folds between ages four and five years. Second, our results describe the common errors children made while folding.The most common errors were folding the edge or tip of the paper to the line (not along the line), over-imitating Fold 1 for Folds 2 and 3, and bending but not actually making a crease in the paper to make a fold. Third, the present results found that children with more age-appropriate problem solving skills attempted more folds. These results and their implications are discussed in more detail below.

Ages of Folding Ability Development

Our results suggest that children as young as 27 months of age were able to make folds along a line. However, the median age was between 4 and 5 1/2 years. These data suggest that while individual differences in the development of paper folding exist, ages 4–6 years might be the timeframe during which folding paper along a line begins to be observed in the majority of children. Therefore, kindergarten may be an important time for paper folding skills to be practiced. Interestingly, most research that has investigated folding practice and origami as an intervention for spatial or math skills has been done with older elementary and middle-school students (4^th-7^th grade). However, the findings of the present study suggest that folding practice as an intervention might be effectively implemented years earlier. Indeed, in school systems outside of the United States, folding and origami practices are implemented in early education (Tobin, Hsueh, & Karasawa, 2009; Whitburn, 2003), and it is possible that interventions targeting spatial skills may have a greater impact if implemented earlier. Future research is needed to examine whether folding interventions in younger children are feasible and are able to enhance skills that are linked to success in STEM disciplines.

In terms of the development of physical paper folding compared to mental paper folding, our finding that physical paper folding is present in the majority of children between 4.5 and 5.5 years of age is consistent with the previous finding that mental paper folding tasks emerged at approximately 5.5 years (Harris et al., 2013). From these ages, it is possible that physical paper folding may developmentally precede mental paper folding. Future research that includes measures of both physical and mental paper folding is needed to be able to examine how these two aspects of paper folding interact during early childhood.

Common Errors in Paper Folding

Understanding the descriptive factors for why a child was unable to complete a fold is a critical step in understanding how folding ability develops. In investigating the error patterns of folding across children, we found that folding the edge (or corner) of the paper to the line and over-imitating Fold 1 were the most frequent errors, followed by bending but not actually making a crease in the paper. Edge-to-line errors were not related to age, whereas bend-not-crease and overimitating-fold-1 errors were most present in the youngest children. Speculatively, some of these errors (e.g., folding the edge or corner to the line rather than along the line, over-imitating folds) may be linked to cognitive planning and attentional issues. Other errors (e.g., scrunching, rolling, ripping), may be linked to motor control issues. Further research that collected measures of executive function in conjunction with independent assessments of motor function is needed to examine these hypotheses.

Factors Underlying Age-Related Changes in Folding Ability

Across the sample of children, the only significant correlation after controlling for age showed that more age-appropriate problem solving related to the number of folds a child attempted. Therefore, declining to fold might have been related to the cognitive contributions to paper folding, such as understanding the instructions or the ability to plan the fold. However, we did not have a standardized, non-parent-report measure of these cognitive abilities, which is a limitation of the present study and a key avenue for future research.

In terms of motor contributions to folding, grip strength and pinch strength covary with age. Thus, it is perhaps unsurprising that controlling for age in these analyses greatly diminished the observed effects of grip and pinch strength on folding performance across the wide age range. Speculatively, the motor contributions to paper folding may become more pronounced at ages when children are better able to understand and follow directions, which might affect how we should interpret folding performance in young children or individuals with lower mental ages when folding is used as an assessment of fine motor ability. While the measure of development and motor skills were far from exhaustive in the present study, these findings lay the groundwork for future investigations to better understand the interaction between cognition, development, and motor skills in children.

Limitations and Future Directions

While the results of this study are important in characterizing age-related differences in paper-folding ability, there were a number of limitations. First, folding is thought to involve visual-spatial skills (Roid & Miller, 1997; Thorndike et al., 1986) and executive function (Anderson, 2002). However, the present study did not include standardized cognitive assessments to evaluate the contribution of these cognitive domains to folding in the young children. This will be a key avenue for future research. Second, the ASQ-3 is most useful for distinguishing typical versus atypical development. Some of the factors that we investigated in this study, such as pinch and grip strength, may be more closely linked in children with atypical development than in children with typical development. There may be some interesting age-related transitions where certain factors serve to constrain or facilitate aspects of motor behavior. If this were true, we would not find a linear relationship, a possibility that needs to be investigated in a future study. Third, only 62% of the parents completed and returned the survey, which led to incomplete parent-report assessments (like the ASQ-3) and demographic information for a number of the children. While there is no evidence to suggest that children with a parent survey were systematically different from those whose parents did not complete the survey, it is nonetheless important for future investigations to include a larger set of data from a more diverse sample. Further, there were different patterns of performance among the three folds, such that performance on Fold 3 was decreased relative to other folds. Fold 3 might have been different because of having to fold a piece of paper that already has folds in it, increased cognitive demands (i.e., something about the nature of this fold is harder), or decreased motivation to complete the third and final fold. Future research should counterbalance the order of the folds in order to determine whether the performance differences among the folds were due to the nature of the fold (corner fold versus long [hotdog] fold), order effects, or the cumulative nature of folding three folds on the same piece of paper.

Conclusion

While paper folding has been used in the past for assessments of visual-spatial reasoning (Youngstrom, Glutting, & Watkinds, 2003) and as part of perceptual motor assessments in children ages four and older (Bruininks & Bruininks, 2005), to our knowledge the current study is the first to document the early emergence of this behavior. Paper folding may not be viewed as important to everyday life in the same way as some other relatively early emerging motor skills (e.g., spoon use, drawing, writing). However, given the importance placed on mental paper folding in tests of cognitive and visual-spatial reasoning, it is important to understand when and how actual paper folding emerges. Follow-up studies are needed to investigate more closely the links between actual and mental paper folding. If von Hofsten (2012) is correct that early motor behavior serves as the foundation for later cognitive skills, we would expect proficient paper folding early in life (as described by the present study) to be a precursor of later performance on mental paper-folding and visual-spatial tasks.