Put your hands up! Gesturing improves preschoolers’ executive function

Journal of Experimental Child Psychology 173 (2018) 41–58

Contents lists available at ScienceDirect

Journal of Experimental Child Psychology journal homepage: www.elsevier.com/locate/jecp

Put your hands up! Gesturing improves preschoolers’ executive function Candace L. Rhoads a,1, Patricia H. Miller a,⇑, Gina O. Jaeger b a b

Department of Psychology, San Francisco State University, San Francisco, CA 94132, United States Department of Human Ecology, University of California, Davis, Davis, CA 95616, United States

a r t i c l e

i n f o

Article history: Received 23 August 2017 Revised 19 March 2018

Keywords: Gesture Executive function Development Preschoolers DCCS Training

a b s t r a c t This study addressed the causal direction of a previously reported relation between preschoolers’ gesturing and their executive functioning on the Dimensional Change Card Sort (DCCS) sorting–switch task. Gesturing the relevant dimension for sorting was induced in a Gesture group through instructions, imitation, and prompts. In contrast, the Control group was instructed to ‘‘think hard” when sorting. Preschoolers (N = 50) performed two DCCS tasks: (a) sort by size and then spatial orientation of two objects and (b) sort by shape and then proximity of the two objects. An examination of performance over trials permitted a fine-grained depiction of patterns of younger and older children in the Gesture and Control conditions. After the relevant dimension was switched, the Gesture group had more accurate sorts than the Control group, particularly among younger children on the second task. Moreover, the amount of gesturing predicted the number of correct sorts among younger children on the second task. The overall association between gesturing and sorting was not reflected at the level of individual trials, perhaps indicating covert gestural representation on some trials or the triggering of a relevant verbal representation by the gesturing. The delayed benefit of gesturing, until the second task, in the younger children may indicate a utilization deficiency. Results are discussed in terms of theories of gesturing and thought. The findings open up a new avenue of research and theorizing about the possible role of gesturing in emerging executive function. Ó 2018 Elsevier Inc. All rights reserved.

⇑ Corresponding author. Fax: +1 415 388 2398. E-mail address: [email protected] (P.H. Miller). Current address: Department of Psychiatry and Behavioral Sciences, University of Washington, Seattle, WA 98195, United States. 1

https://doi.org/10.1016/j.jecp.2018.03.010 0022-0965/Ó 2018 Elsevier Inc. All rights reserved.

42

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

Introduction During early childhood, action and cognition are closely linked (e.g., Kontra, Goldin-Meadow, & Beilock, 2012). One type of bodily movement—gesturing—appears to be particularly important for children’s cognition. Gesturing both reflects knowledge that cannot yet be verbalized and changes thinking (Goldin-Meadow & Alibali, 2013). Gesturing helps children’s learning and problem solving on a variety of tasks such as recall (Cameron & Xu, 2011), math (Broaders, Cook, Mitchell, & GoldinMeadow, 2007), and spatial reasoning (Ehrlich, Levine, & Goldin-Meadow, 2006). Evidence that gesturing can be for one’s own thinking, rather than merely for communicating with others, is that congenitally blind individuals spontaneously gesture during reasoning tasks (Iverson & Goldin-Meadow, 2001). Gestures may contribute to both cognitive ability and performance (effective use of ability). Regarding cognitive ability, gestures may contribute to thinking by representing information not found in verbal representations such as holistic visual imagery (Goldin-Meadow & Alibali, 2013). Sometimes this additional information represents knowledge that a child cannot yet express in words. For example, research examining children’s spontaneous gestures while explaining how they solved a balance beam task revealed that one third of the children’s gestures conveyed information that did not match their speech such as gesturing about weight while discussing distance from the fulcrum (Pine, Lufkin, Kirk, & Messer, 2007). Further analysis of these mismatches indicated that on 80% of instances, children’s gestures contained more advanced knowledge than their accompanying verbalizations, and that on more than half of these instances, information was conveyed in gesture before—if ever—showing up in speech. In general, information was also clearer and more precise in gestures than in speech. Thus, children often know more than they can articulate, and gestured information can run ahead of verbalized information. This implicit gestural knowledge may become explicit knowledge (Broaders et al., 2007). Gesturing is also an effective form of instruction. For example, previous research found that instructing third and fourth graders to gesture a correct math problem-solving strategy led to greater learning than when instruction involved no gesturing or partially correct gestures (Goldin-Meadow, Cook, & Mitchell, 2009). Thus, it would appear that gesturing can activate implicit ideas that then facilitate learning. Gestures may not only improve cognitive abilities but also facilitate children’s application of their cognitive abilities during problem solving. In particular, gesture is thought to lighten the cognitive load during problem solving (e.g., Goldin-Meadow, Nusbaum, Kelly, & Wagner, 2001; Wagner, Nusbaum, & Goldin-Meadow, 2004). For example, while testing children on math equivalence problems, Goldin-Meadow and colleagues (Goldin-Meadow et al., 2001) increased children’s cognitive load by giving them a list of letters to remember. Children who spontaneously gestured while providing explanations for the math problems recalled more items than those who did not, which was taken as evidence that gesturing on the explanation trials reduced cognitive load, allowing children to allocate more cognitive resources toward word recall. Given the considerable evidence that gestures can facilitate cognitive ability and performance, it is surprising that researchers have only recently begun to consider the possible role of gesturing in one of the most active areas of research on cognitive development—children’s executive function (EF). Executive function refers to cognitive processes used to control thinking and behavior when trying to achieve a goal, particularly when solving a novel problem. EF is related to various important outcomes in children, including academic achievement, behavior in the classroom, and social competencies (Best, Miller, & Jones, 2009). Studies have identified several processes underlying the development of EF such as brain maturation (Johnson, Munro, & Bunge, 2014), the ability to construct a hierarchical set of rules (Marcovitch & Zelazo, 2009), physical activity (Davis et al., 2011), and parenting behaviors (e.g., Deater-Deckard, 2014). The purpose of the current study was to examine whether gesturing can facilitate children’s performance on an EF task. So far, only one study has examined gesturing in an EF task in young children. O’Neill and Miller (2013) demonstrated that preschoolers who spontaneously gestured the relevant dimension on an

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

43

EF sorting task (e.g., sorting by shape) were more likely to sort accurately than children who were less likely to gesture. In fact, gesturing appeared to be more important than age in predicting children’s EF. Given this correlational relation between gesturing and EF, the next step is to determine whether this relation is causal. Thus, we asked whether instructing children to gesture improves EF compared with a control group with no such instruction. The current study used the Dimensional Change Card Sort (DCCS) task (Zelazo, 2006) used by O’Neill and Miller (2013), in part because it often serves as an assessment of EF in young children and, thus, much is known about young children’s performance on this task. Moreover, this conflicting rules task is one of two EF assessments in the new NIH Toolbox for researchers (National Institutes of Health, 2016). In the DCCS, a child first sorts cards with line drawings on the basis of one dimension (e.g., color), but then must switch to sorting on the basis of the other (previously irrelevant) dimension (e.g., shape), which is thought to draw on EF. Although 3-year-olds usually find it easy to sort on the first dimension, they tend to perseverate on this first rule after the task switches to the second rule. By 5 years of age, children usually can make the switch. The DCCS likely draws on all three basic components thought to constitute EF: inhibition, working memory, and cognitive shifting (Miyake et al., 2000). The task requires inhibition to suppress processing of the irrelevant dimension, working memory to maintain representations of the relevant task rules, and task switching for updating these processes after the rule switch (Wiebe, Morton, Buss, & Spencer, 2014). Gestures may facilitate EF performance on this task by scaffolding existing EF components or by improving EF ability per se. Regarding the former, with respect to the working memory component there is evidence that adding gesturing to speaking frees up working memory resources (Ping & Goldin-Meadow, 2010). On the DCCS task, gesturing could carry some of the cognitive load, thereby helping children to keep track of which dimension is relevant and which is irrelevant, and update this information as needed in working memory. Gesturing the relevant dimension could also aid the inhibition and shifting components by reducing the pull of the prepotent—but now irrelevant—dimension and reduce perseveration by strengthening the representation of the newly relevant dimension. Thus, gesturing may improve children’s performance on the DCCS task by helping children to apply their existing EF abilities. Regarding possible improvement of EF per se, gesturing may change how information is represented, updated, and monitored in working memory, thereby encouraging reflection about possible responses on EF tasks (Marcovitch & Zelazo, 2009). More specifically, gesturing the relevant dimensions may make children’s conflict more salient, which in turn could encourage reflection and lead to the construction of a higher-order rule that resolves the conflict. In addition, as in previous gesture research (e.g., GoldinMeadow & Alibali, 2013), gesturing may change children’s thinking by encoding and representing important information not yet included in their verbal representations in working memory. In the current study, on each trial preschoolers were prompted to gesture the relevant dimension. One thorny problem when instructing preschoolers to use a particular strategy is the danger that the task of remembering to use the strategy and how to use it adds to children’s cognitive load and also may distract the children from the problem-solving task itself, which could actually worsen task performance. In our case, remembering to gesture could overtax preschoolers’ limited cognitive capacity and, thus, negate any benefit from increased gesturing. We addressed this issue by encouraging gesturing but not requiring it and by keeping our gesture instructions somewhat implicit, and thus less resource demanding, through suggestion, modeling, and practice. This Gesture group was compared with a Control group that was not instructed to gesture. Instead, the Control group was told to ‘‘think hard” about the materials. This wording was intended to increase attention to the materials and the task so that any unique effect of gesturing, beyond increasing overall attention to the materials, might be detected. In addition, this wording made the amount of time spent giving the instructions more similar. Although a few studies with a gesture instruction condition have included an additional condition in which school-aged children were told not to gesture (e.g., Goldin-Meadow et al., 2009), this procedure is problematic for preschoolers. First is the problem of increasing cognitive load (discussed earlier) due to children constantly reminding themselves not to gesture, remembering to keep their hands in their lap, and so on. Second, telling young children not to gesture actually could increase their thinking about gesturing. According to ironic process theory (Wegner, 1989), people told not to think about a particular

44

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

topic will think about it more than people allowed to think freely. Finally, O’Neill and Miller (2013) found that children produced few spontaneous overt gestures (although they could gesture the basis of their sorting when requested), and so there would be few spontaneous gestures to reduce. For these reasons, we decided not to include a condition in which children were told not to gesture. The effects of gesturing on cognitive performance may be limited to the right developmental moment—either in a macro sense where the effect occurs only during certain developmental phases (as suggested by age differences) or in a micro sense where any benefit emerges microgenetically over trials. Regarding the former, we included ages 3–5 years because this is when children move from poor to good performance on the DCCS and, thus, the ages when gesturing is more likely to affect performance. It is also the age span for which previous research (O’Neill & Miller, 2013) found an association between gesturing and sorting accuracy on the DCCS. Given that many 4- and 5-year-olds already are beginning to sort well on the DCCS task, the benefit of gesturing likely is greater in the younger children, whose emerging EF skills need support. As evidence, O’Neill and Miller (2013) found that the association between spontaneous gesturing and accurate sorting was stronger for younger preschoolers. As for microgenetic change, analyses of cognitive performance over trials sometimes show a delayed benefit of an emerging or experimentally introduced skill such as a strategy (e.g., DeMarieDreblow & Miller, 1988). For gesturing specifically, guiding children’s movements during problem solving can have a delayed effect on learning and performance. Brooks and Goldin-Meadow (2016) taught children gestures that were either relevant or irrelevant to solving mathematical equivalence problems. Although neither group showed increased understanding of these problems on tasks immediately following gesture training, after receiving subsequent verbal instruction on these problems, children in the gesture-relevant group improved their understanding of equivalence more than children who had been taught irrelevant gestures. Thus, in our study we explored both age and temporal (over trials and over two tasks) variables. We included multiple ways of detecting children’s use of gesture. In addition to observations of overt gesturing, we included probes for verbal and gestural explanations at the end of each of two EF tasks. As discussed earlier, children tend to convey different information through their speech and gestures. In summary, the current study compared a group of children instructed to gesture with a control group instructed to ‘‘think hard,” with no mention of gestures. More accurate sorting in the Gesture group would suggest that gestures facilitate EF performance; no condition differences would suggest that gestures provide no unique benefit beyond perhaps increasing attention to the task. We examined gesturing as a predictor of EF within two age groups because gesture may be more useful during certain developmental periods. If gesture facilitates EF only when EF is emerging, then gesturing should help younger children more than older children. If elicited gestures require time or experience to have an effect, then improvement in the Gesture condition should not appear until late in the task. Accordingly, we looked at (a) relationships among gesturing, age, and sorting accuracy and (b) patterns of improved accuracy over trials. Such an analysis of patterns of change in accuracy (and gesturing) also might provide clues to the underlying processes involved if gesturing does in fact facilitate EF in children. No differences between ages or task would suggest the lack of importance of these variables for theoretical explanations of the role of gestures in EF. We assessed both children’s use of gestures during the DCCS task and their ability, when asked, to provide a gesture that explains the basis of their sorting (i.e., indicates the dimension on which they sorted). If the gesture probe elicited more gesturing in children in the Gesture condition than in the Control condition, this would provide additional evidence that children in the Gesture group were using gestures in the service of their sorting.

Method Participants Participants were 50 children aged 30–60 months (M = 44 months, SD = 7.94; 28 boys and 22 girls) at four day-care centers. An additional 2 children who were unable to successfully complete the task

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

45

due to inattention and failure to follow instructions were dropped from the sample. The 50 children achieved the criterion of sorting correctly on at least 9 of the 12 pre-switch trials (p = .054, binomial distribution) over the two tasks, and all but 1 of these children sorted correctly on at least 5 of the 6 pre-switch trials on both tasks. The sample was racially and economically mixed but primarily middle class. Parental written consent and child verbal assent were obtained. Children were tested individually at their day-care center. Materials A widely used assessment of preschoolers’ EF, the Dimension Change Card Sort (Zelazo, 2006), was used to measure EF. In this task, children sort cards with drawings according to one dimension for several trials and then are asked to sort according to a different dimension. The switch from one sorting rule to another is thought to draw on EF. The standard DCCS task uses the dimensions of color and shape, but instead we used dimensions more easily represented in gestures for our two tasks. One task, developed by O’Neill and Miller (2013), used yellow bears that could be sorted by size or spatial orientation. The other task, developed for this study, used red drinking straws that could be sorted by shape (straight or ‘‘squiggly”) or proximity (distance between the two straws). Each task used two target cards (one on each tray) and 12 test cards (7.6  12.7 cm) to generate 12 trials (see Fig. 1 for examples of the size/spatial and shape/proximity task test cards). On each trial, children sorted the test card by selecting one tray (each 12.7  17.8  3.8 cm) in which to place the card. Each tray held a target card showing the dimension to be used for sorting. On the size/orientation task, one target card displayed a large yellow bear sitting upright and one target card displayed a small yellow bear rotated 90 degrees. When size is the relevant dimension, all test cards with large bears, regardless of orientation, should be placed in the tray with the large bear and all test cards with small bears, regardless of orientation, should be placed in the tray with the small bear. When orientation is the relevant dimension, all test cards with an upright bear, regardless of size, should be placed in the tray with the upright bear, and so on. The shape/proximity target cards displayed two straight straws that were close together or two squiggly straws that were far apart. The test cards to be sorted were, for size/orientation, a big bear turned 90 degrees clockwise or a little bear sitting upright and, for shape/proximity, two straight straws far apart or two squiggly straws close together. Procedure Children were randomly assigned to either a Control condition (Mage = 3; 8 [years; months], SD = 7. 47 months; n = 24) or a Gesture condition (Mage = 3; 9, SD = 8.46 months; n = 26) and then to one of two task orders: size/orientation followed by shape/proximity or the reverse. The orders had similar gender and age ratios. A small video camera 3 feet from the children recorded the sessions.

Fig. 1. The eight types of test cards used for size/spatial (bears) and shape/proximity (straws) tasks.

46

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

The experimenter followed a modified script of the DCCS (Zelazo, 2006). In the Gesture condition, for gesturing shape, the experimenter moved her flat hand up and down to indicate straight or moved it around and around in a corkscrew pattern to indicate squiggly. For gesturing proximity, she placed her hands with palms near to or far from each other. For gesturing size, her curved hands facing each other created a small or large space. For gesturing orientation, her finger pointed up or was rotated 90 degrees clockwise. In the Gesture condition, the experimenter began the session by labeling the target cards (e.g., ‘‘a big bear sitting up”). She then described the game (e.g., ‘‘the size game”) and stated the dimension on which they should sort (e.g., ‘‘all the big bears [gestured big size] go here, and all the little bears [gestured small size] go here”). Then, she said, ‘‘I like to think with my hands: big bear [gestured big] and little bear [gestured small]. See, here’s a big bear. So, it goes here [sorted test card into big bear tray]. If it’s a big bear [gestured large], it goes here [pointed]. If it’s a little bear (gestured small], it goes here [pointed].” She then repeated the rule for sorting by size, held up a new test card, and said, ‘‘Now, here is a little one. Can you think with your hands? [gestured big and little while saying this and looked expectantly at the child].” If the child did not gesture, the experimenter gestured again and looked expectantly again. She then said, ‘‘Can you point to where the little one goes?” Regardless of whether the child was correct or incorrect, the experimenter sorted the test card into the correct box. If the sort was correct, the experimenter praised the child; if it was incorrect, she corrected the child and explained why she (the experimenter) sorted the card where she did. Then, Trial 1 began with the experimenter saying, ‘‘Now it’s your turn to think with your hands [lifted hands and paused and looked expectantly until the child raised hands]. Remember, if it’s big [gestured big] it goes here, and if it’s little [gestured little] it goes here. And remember to think with your hands.” She then gave the child the first test card and said and gestured the relevant dimension. The child pointed to one of the trays. On each trial, the experimenter repeated the rule and said and gestured the relevant dimension. After the 6 pre-switch trials, the child was told that they were going to play a new game (e.g., ‘‘the upright and on its side game”). The experimenter repeated the above procedure, wording, gesturing, and so on. After 6 post-switch sorts, the experimenter asked about the basis of sorting on Trial 12 with two probes: ‘‘Why did you put that card in that box?” and ‘‘Can you show me with your hands why you put that card in the box?” Next, the second task began with a new set of cards, with new dimensions, to sort. The procedure for the first task was repeated. The Gesture and Control conditions were identical except that in the Control condition the experimenter substituted ‘‘thinking hard” for every time the experimenter in the Gesture condition mentioned thinking with hands or gestured a dimension. For example, when introducing the size game, the experimenter said, ‘‘I like to think hard: big bear [experimenter looked thoughtful] and little bear [experimenter looked thoughtful].” This ensured equal attention to the task in the two conditions. Coding of gestures Gestures and verbalizations were coded, following O’Neill and Miller’s (2013) modification of Krauss, Chen, and Chawla’s (1996) typology of gestures (Fig. 2), for all 24 trials and the probes at the end of each task. They were coded as relevant (referred to the relevant dimension for sorting) or not, and relevant gestures were coded as lexical or deictic. Lexical gestures are defined as hand movements that typically co-occur with speech and appear to be related to the semantic content of the speech they accompany (e.g., tracing a circular shape in the air to depict roundness), whereas deictic gestures are defined as hand movements used to request or declare something such as pointing to or holding an object up in display (Bates, Camaioni, & Volterra, 1975; Krauss et al., 1996). For example, in the current study, a common lexical gesture for orientation was holding out one hand with the palm facing up and rotating it 90 degrees. A common deictic gesture for orientation was pointing to the top of the bear’s head for both the vertical and horizontal bears. Both lexical and deictic gestures needed to convey the relevant dimension for sorting. Examples of gestures in each coding category appear in Fig. 2. A second rater, blind to the purpose of the study, coded the entire set of trials and probes for a subset of children that constituted 27% of the data (112 gesture and verbal responses). Interrater reliability for the six categories was good (kappa = .92, p < .001), with 96% agreement between the raters.

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

47

Fig. 2. Gesture coding scheme adapted from Krauss et al. (1996) and O’Neill and Miller (2013). Categories of relevant gestural and verbal explanations are in bold font.

Results After presenting some preliminary analyses, we address the relations between gesturing and number of correct sorts and then look at patterns of gesturing and correct sorting over trials by age and condition. Finally, we examine children’s gesturing and verbalizations during explanations of their sorting.

Preliminary analyses An analysis of variance (ANOVA) examined whether two variables not expected to have an effect— gender and task order—were in fact not significant. Following typical practice in studies of the DCCS, because performance typically is near perfect on the pre-switch trials, we examined the 6 post-switch trials on the two tasks (i.e., 12 trials total). A 2  2  2  2 factorial ANOVA examined effects of age (mean split: younger, M = 3; 2, SD = 3.75 months vs. older, M = 4; 3, SD = 4.99 months), condition (Gesture vs. Control), gender, and task order (bears then straws vs. straws then bears) on sorting accuracy. There were significant main effects of age, F(1, 34) = 15.02, p < .001, g2p = .31, and condition, F(1, 34) = 7.25, p = .01, g2p = .18, but not Age  Condition. Older children had more accurate sorts (M = 8.04) than younger children (M = 3.23). As predicted, children in the Gesture condition had more accurate sorts (M = 6.62) than children in the Control condition (M = 4.38). There were no main effects of gender, F(1, 34) = 1.68, p > .05, or task order, F(1, 34) = 0.03, p > .05. A significant Gender  Task Order interaction, F (1, 34) = 6.40, p = .02, g2p = .16, indicated that when the bears task was first, boys and girls performed

48

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

similarly (mean correct sorts: boys = 4.60, girls = 6.17), but when the straws task was first, boys (M = 7.60) performed better than girls (M = 2.80). This interaction is uninterpretable and, importantly, did not interact with condition or age, so it was not examined further. Next, we checked whether any differences in accuracy between the Gesture and Control conditions were not simply due to differences in how long children took to respond. Response time was defined as the time between the moment the experimenter presented the test card to be sorted and the moment it was placed in the bin. Although the average response time was slightly, but significantly, greater for the Gesture condition (M = 8.32 s, SD = 1.84) than for the Control condition (M = 6.08 s, SD = 1.13), t(1, 48) = 5.12, p < .001, response time was not related to performance: The bivariate correlation between time and sorting accuracy was not significant, r (50) = .07, p = .62. Thus, any condition difference in sorting accuracy is unlikely to be due to group differences in response time. A 2 (Age)  2 (Condition) ANOVA provided a manipulation check on whether instructing children in the Gesture condition to gesture actually led to their gesturing more than children in the Control condition. Both pre- and post-switch trials were included because gesture instructions were given during both phases. Age and condition were independent variables, and number of relevant gestures was the outcome variable. There was a significant effect of condition, F(1, 46) = 7.29, p = .01, g2p = .14, with children in the Gesture condition producing significantly more gestures on the sorting trials (M = 5.36, SD = 6.20) than children in the Control condition (M = 0.97, SD = 4.89). There were no effects of age, F(1, 46 = .19, p > .05, or Age  Condition, F(1, 46) = 0.42, p > .05. Thus, across both age groups, the instructions to gesture in the Gesture condition successfully induced children to produce more gestures than children in the Control condition. Notably, all gestures produced by the Gesture group on the 24 trials were relevant (vs. 72% for the Control group). The Gesture group produced 136 lexical and 8 deictic gestures, and the Control group produced 0 lexical and 25 deictic gestures. Older children produced 56 lexical and 24 deictic gestures, and younger children produced 80 lexical and 9 deictic gestures. Children produced 73 lexical and 18 deictic gestures on their first task and produced 63 lexical and 15 deictic gestures on their second task. In the Gesture condition, all children produced relevant gestures during the initial instructions on both tasks. On post-shift sorting trials, all but 2 children in the Gesture condition gestured at least once on the first task, and all but 1 child in the Gesture condition gestured at least once on the second task. Across all sorting trials, 69% gestured at least once and 8% gestured on at least 20 of these 26 trials. Few children in either condition ever spontaneously produced a relevant verbalization on the sorting trials (e.g., saying ‘‘big”)—1.3% of the trials—and so these few verbalizations were not analyzed further. Children produced 7 irrelevant gestures and 43 irrelevant verbalizations. Relations between gesturing and sorting accuracy The first question was whether children in the Gesture condition sorted more accurately after the switch than children in the Control condition. An ANOVA, with post-switch accuracy as the dependent variable, examined condition and age. A significant effect of condition, F(1, 46) = 5.93, p = .02, g2p = .11, indicated that children in the Gesture condition sorted correctly on significantly more trials (M = 6.62) than children in the Control condition (M = 4.38). A significant effect of age, F(1, 46) = 15.67, p < .001, g2p = .25, indicated that older children sorted correctly on significantly more trials (M = 8.04) than younger children (M = 3.23). There was no significant interaction of age and condition. It was important to examine each task separately because gesture prompts continued across trials and, thus, might not have an effect until later trials. Given the statistical problems typically associated with repeated measures on unequal ns (n = 26 in the Gesture condition and n = 24 in the Control condition), we did not include a task (first vs. second) variable in the above ANOVA. Instead, to examine the first and second tasks separately, we conducted an ANOVA with this variable on each of the four age–condition groups. Because the homogeneity of variance assumption was not met for these data, we used Welch’s adjusted F ratios when needed, which required us to conduct one-way ANOVAs for each age separately. A one-way ANOVA, with post-switch accuracy as the dependent variable, showed a significant effect of condition (Gesture > Control) only for younger children on the second task, Welch’s F(1, 21.95) = 6.94, p = .02, estimated x2 = .11. On the other three ANOVAs, test results for younger children on Task 1 were F(1, 24) = 0.26, p = .62, for older children on Task 1 were

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

49

Welch’s F(1, 21.91) = 3.14, p = .09, and for older children on Task 2 were F(1, 22) = 1.67, p = .21. Fig. 3 displays the overall pattern of results, and Table 1 shows the means. Thus, the gesture training significantly affected only younger children during the later phases of sorting. Interestingly, the greater effect of gesturing on the second task than on the first task was not due to younger children in the Gesture condition producing significantly more gestures on the post-switch trials of the second task (M = 0.93, SD = 1.44 on Task 1 vs. M = 1.73, SD = 1.98 on Task 2), t(14) = 1.57, p = .14, a point we return to in the Discussion. We examined the complex relationship among gesturing, age, task, and sorting in more depth by analyzing whether children who gestured more also benefitted more. We conducted a hierarchical regression on each task (first and second), with age and number of relevant gestures as the predictor variables and number of correct sorts as the outcome variable. Children from the two conditions were combined because there was some spontaneous gesturing in the control group, and all trials were included because gestures were prompted on both pre- and posttest trials. A three-step model was performed on each task with (a) age (in months) in the first block, (b) number of relevant gestures during the 12 trials in the second block, and (c) an interaction between age and gesture in the third block. The predictor variables were centered to the mean to avoid multicollinearity. Variance inflation factors (VIFs) and tolerance levels were good (tolerance > .78). For the first task, the overall model was significant, F(3, 46) = 5.32 p = .003. Age, entered at Step 1, was significant, predicting 24.1% of the variance in number of correct sorts (R2 = .241), F(1, 48) = 15.21, p < .001. Entered at Step 2, gesture predicted an additional 1.6% of the variance (DR2 = .016), F(1, 47) = 1.00, p = .32. Entered at Step 3, the interaction variable predicted 0.1% (DR2 = .001) of the variance, F(1, 46) = 0.08, p = .78. Thus, on the first task, after controlling for age, the number of relevant gestures and the interaction variable did not predict accuracy. On the second task, the overall model was significant, F(3, 46) = 8.27, p < .001. Using the same predictors and sequential method of entry, age predicted 24% of the variance in accuracy, which was significant (R2 = .240), F(1, 48) = 15.17, p < .001. At Step 2, relevant gestures predicted an additional 4.6% (DR2 = .046) of variance in accuracy, which approached significance, F(1, 47) = 3.04, p = .09. At Step 3, the Age  Gesture interaction variable predicted 6.4% (DR2 = .064) of the variance, which was significant, F(1, 46) = 4.55, p = .04. Thus, as suggested by the earlier ANOVAs and Fig. 3, these two regressions show that the greatest impact of number of gestures produced was in the second task, especially for younger children. In summary, the ANOVAs showed that the younger children in the Gesture group sorted more accurately than the Control group by the second task, and regression analyses showed that greater gesturing was associated with more accurate sorts in young children on the second task.

Mean number of cards sorted correctly

6

5

4 Older Gesture 3

Older Control Younger Gesture Younger Control

2

1

0 1

2 Task

Fig. 3. Number of cards sorted correctly during post-switch trials of first and second tasks. Error bars represent standard errors.

50

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

Table 1 Mean number of post-switch trials sorted correctly (out of 6 on each task). Group

Older gesture Older control Younger gesture Younger control Total

Task 1

Task 2

Total

M

SD

M

SD

M

SD

4.82 2.85 1.53 1.09 2.50

2.40 3.05 2.29 2.07 2.79

4.82 3.46 2.87 0.55 2.94

2.40 2.70 2.92 1.51 2.83

9.64 6.31 4.40 1.64 5.44

3.96 4.92 4.82 3.35 5.08

Finally, only relevant gestures had an impact. The correlation between number of irrelevant gestures and number of correct sorts was nonsignificant, r(50) = .10, p = .50. Patterns over trials To find clues about underlying developmental processes, we examined gesturing and sorting accuracy over trials. Figs. 4 and 5 illustrate the percentage of children who sorted correctly and produced a relevant gesture by trial, respectively. Examining the pattern of performance across trials on each figure and comparing the two figures clarifies group differences as well as the relations between gesturing and sorting accuracy. The advantage for the Gesture group comes after the switch. All four groups sorted very accurately on the pre-switch trials but then differentiated after the switch from one rule to another (Trials 1–6 vs. 7–12 and Trials 13–18 vs. 19–24). Specifically, only the older Gesture group continued to sort accurately. Being younger or being in the Control group was associated with deteriorated sorting after the switch, considered a marker for poor EF on the DCCS task. The Gesture group had more accurate sorts than the Control group for both ages and tasks except for the younger children on the first task. In fact, on the final 3 trials of the last task, the sorting of younger children in the Gesture group was virtually the same as that of older children in the Control group, suggesting that gesturing can override most of the effects of age on the DCCS task. Regarding gesturing and sorting across trials, the most striking pattern is the lack of correspondence between gesturing (Fig. 5) and accurate sorting (Fig. 4) at the level of individual trials. For

Fig. 4. Percentage of children who sorted accurately per trial.

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

51

Fig. 5. Percentage of children producing a relevant gesture for each trial during the DCCS.

example, although children in the Gesture condition usually showed the most gesturing on the first trial of each set of 6 trials, this never led to more accurate sorts on the first trial. Similarly, even though gesturing usually declined over trials within a trial set, sorting did not show that pattern. Thus, the gesture manipulation benefitted sorting overall rather than at the level of individual trials. However, the correspondence appears to be better, in the younger group, when individual children’s gesturing and sorting are considered. Among younger children in the Gesture condition who actually gestured on a given trial, 50% sorted correctly on that trial compared with 33% for children in the Gesture condition who did not gesture on that trial. A similar comparison for the older children in the Gesture group showed no difference (79% sorted correctly when they gestured and 81% sorted correctly when they did not).

Verbal and gesture probes Additional evidence of the impact of gesture training came from the children’s explanations. At the end of each task, children were asked why they put the card where they did and then whether they could explain this with their hands. Table 2 presents the percentage of children giving a relevant explanation on each probe, specifically, a relevant verbal response on verbal probes or a relevant gesture on gesture probes. The percentages in Table 2 show that the Gesture group produced more relevant answers than the Control group in each age group on each of the probes except for the verbal responses of the younger children. Chi-square values (with ages combined to increase the N—26 in the Gesture condition and 24 in the Control condition) were significant on the second probe (gesture), v2(1, N = 50) = 5.98, p = .01, and on the fourth probe (gesture), v2(1, N = 50) = 5.13, p = .02, and were borderline significant on the first probe (verbal), v2(1, N = 50) = 3.57, p = .06. The conditions did not differ on the third probe (verbal), v2(1, N = 50) = 0.55, p > .05, likely because the younger children gave few relevant explanations in either condition. Thus, for the most part, the Gesture group produced more verbal and gestural explanations than the Control group, whose members produced very few relevant responses. Of particular interest is the large number of relevant verbal explanations produced by the older children in the Gesture group—approximately twice as many as the older children in the Control group (64% vs. 31% on the first verbal probe, 73% vs. 38% on the second verbal probe)—even

52

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

Table 2 Percentage of children in each group giving a relevant response on each probe. Gesture group

Older gesture Older control Younger gesture Younger control

Task 1

Task 2

Verbal

Gesture

Verbal

Gesture

63.6 30.8 33.3 9.1

45.5 7.7 20.0 0.0

72.7 38.5 6.7 9.1

18.2 0.0 20.0 0.0

though the children in the Gesture group were given gesture training, not verbal training (beyond the verbal instructions also given to the children in the Control group). Thus, although gesture training did not significantly improve older children’s sorting, it appears to have increased their awareness of the basis of their sorting. In these probe analyses, we included all the children regardless of whether they sorted correctly or incorrectly on the two trials with the probes. That decision was based partly on the fact that 23% of the children in the Gesture condition gestured the relevant attribute in one or both of the gesture probes even though they had sorted according to the irrelevant attribute on those trials (e.g., incorrectly sorted by size but gave a relevant explanation based on orientation). Still, one could argue that only children with correct sorts should be examined given that children would be unlikely to give an explanation referring to the correct dimension if they had sorted on the basis of an incorrect dimension. Thus, we also conducted an analysis that included only the probe trials on which children sorted correctly. Because this reduced the number of younger children to 2, only the older children could be examined. A Fisher’s exact test on each probe examined whether the gesture group had a greater percentage of relevant answers compared with the control group. Although the Gesture group produced more relevant answers than the Control group on all four comparisons (67% vs. 57%, 44% vs. 14%, 88% vs. 57%, and 13% vs. 0%), on no probe did the difference reach significance (all chi-square ps > .05), likely due to the reduced number of children. Discussion Adults use gesturing during problem solving, especially during difficult problems (Chu & Kita, 2011; Morsella & Krauss, 2004). The current study suggests that gestures help even preschoolers, especially when the task becomes difficult after a switch in a sorting rule, which calls on EF. We first discuss how gestures might enhance young children’s EF. We then address changes in performance across trials that suggest clues as to the process by which gestures become effective and discuss a utilization deficiency and covert gesture-generated representations. We next address alternative explanations of the results. We conclude by highlighting how the results advance our understanding of the development of both EF and gesturing as well as their developmental relations. Relations between gesturing and EF Building on the earlier observed correlation between preschoolers’ gesturing and their EF (O’Neill & Miller, 2013), this study provides the first evidence supporting a causal relationship between the two. When preschoolers—especially younger ones—were induced to gesture, they showed improved sorting accuracy on a sorting switch EF task compared with control children told to ‘‘think hard.” In fact, by the second task, gesturing overcame the age difference in EF such that younger children in the Gesture condition performed much like older children in the Control condition. Several analyses provide converging evidence for these conclusions. First, ANOVAs showed that younger preschoolers in the Gesture condition sorted more accurately after the switch than those in the Control condition. Second, a regression analysis showed that, among younger children on the second task, greater gesturing was related to more correct sorts. Third, the trial-by-trial analysis showed that gesturing had its effect after the switch, which is when EF comes into play. The fact that

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

53

gesturing mainly helped the younger children suggests that gesturing is particularly helpful during the earlier phases of development of EF involved in the DCCS task. Notably, gestures improved EF despite the fact that even our somewhat implicit instructions carry some cognitive load (i.e., remembering to gesture), which could have decreased EF. In addition, adding gestures to the pre-switch trials in fact could have worked against performance, rather than for performance, by increasing the salience of that dimension and, thus, making the switch to the other dimension even more difficult. Gesturing could facilitate EF performance, or even improve EF abilities, in several ways. Although the current study does not tease apart the contributions to EF ability versus performance on the DCCS task, theories of EF development suggest how both ability and performance could be affected in the current study. Regarding facilitating the use of preexisting EF abilities, Goldin-Meadow and colleagues have proposed (e.g., Goldin-Meadow & Alibali, 2013) that gesturing may take on some of the cognitive load during problem solving. For example, if hand movements keep alive the rule concerning which dimension to sort by, then children might not need to draw on this verbal information in working memory, a component of EF. Young children may find verbal representations of the task taxing because they are still developing their verbal abilities. In addition, gesturing the relevant dimension for sorting after the switch may reduce the effort needed for inhibition of sorting on the basis of the previously relevant dimension. Gestures may also facilitate the application of current EF skills through enhancing sustained attention to the relevant dimension before and after the switch. How long preschoolers can sustain attention to one dimension across trials on one task predicts their ability to sort accurately on post-switch trials on the DCCS task, suggesting similar attentional processes on the two tasks (Benitez, Vales, Hanania, & Smith, 2017). Gesturing may help children to sustain attention on a single source of information and, thus, inhibit the pull of the irrelevant dimension on the DCCS task. Finally, the ‘‘attentional inertia” approach (Kirkham & Diamond, 2003) would emphasize how gestures might help children to inhibit the tendency to perseverate on the first dimension by breaking up the routine of the sorting procedure and, thus, disrupting the continued attention to the now irrelevant dimension after the switch. Of course, several of these processes together may facilitate the application of a child’s EF skills. Regarding the possibility that gesture improves EF abilities per se, rather than only their effective use, gesturing may enhance the representation of the task goal in working memory after the switch by adding a gestural (motoric/spatial) representation to the verbal representation, thereby resulting in a dual representation of the task goal. In one study (Wakefield, Congdon, Novack, Goldin-Meadow, & James, 2014, cited in Goldin-Meadow, 2015), when children learned how to solve a mathematical problem through both verbalizations and self-produced gesture, and subsequently solved comparable problems in a functional magnetic resonance imaging (fMRI) scanner without gesturing, they showed greater activation in a frontal–parietal sensorimotor network than children who had learned only through verbalizations. This outcome suggests that gesture may facilitate learning through establishing sensorimotor representations of the task that are reactivated during later problem solving. Sometimes, gesture may contribute additional information (e.g., holistic/visual) not in the verbal representation, which would produce a richer overall representation of the post-switch sorting rule by tapping into implicit knowledge that is inaccessible to speech (Boncoddo, Dixon, & Kelley, 2010; Broaders et al., 2007). Or, gesturing may activate semantic representations (e.g., Morsella & Krauss, 2004). More radically, gestures may structure, transform, guide, or reorganize cognitive processing, perhaps by highlighting and structuring the relevant dimension for sorting (e.g., Alibali, Spencer, Knox, & Kita, 2011). For example, the meaning conveyed through gestures could provide a framework to support, perhaps through chunking in working memory, representations underlying speech (Wagner et al., 2004). From the perspective of cognitive complexity and control theory and the hierarchical competing systems model (Marcovitch & Zelazo, 2009), gesturing could improve EF ability by stimulating reflection on the task rules. That is, gesturing the relevant dimension may help children to represent the relevant rule (e.g., ‘‘sort by shape”) in working memory and embed it within a higher-order rule (e.g., ‘‘If I sort by shape, then the straws go here; if I sort by distance, then they go there”). In this way, gesturing may transform thinking by encouraging higher levels of consciousness in EF.

54

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

Several of these theories suggest that gesture may work along with speech or verbal representations to improve EF or its application. As evidence that the induced gesturing had some effect on verbal representations, the Gesture group produced more verbal explanations for their sort than the Control group during the probes. It appears that gesturing is somehow affecting verbal representations (e.g., triggering them, strengthening them), or at least children can translate their gesture-based knowledge into explicit verbalizations. Supporting a developmental pathway from gesturing to language to EF, a longitudinal study (Kuhn, Willoughby, Wilbourn, Vernon-Feagans, & Blair, 2014) found that communicative gestures at 15 months of age predicted better EF at 4 years, with the relation entirely mediated by language at 2 and 3 years. In addition, during problem solving, older preschoolers express causal events with gestures before they explain them in sentences (Göksun, Hirsh-Pasek, & Golinkoff, 2010). Notably, during the sorting tasks, children spontaneously produced few relevant verbalizations (on only 1.3% of the trials), suggesting the methodological importance of adding probes for explanations rather than relying only on spontaneous gesturing and verbalizations in studies of children’s gesturing and verbalizations. In particular, even though the difference in sorting accuracy of older children in the Gesture and Control groups did not reach significance, the older children in the Gesture group produced many more explanations—both gestural and verbal—than the older Control group. Thus, the gesture training seems to have enhanced older children’s awareness of the basis of their sorting. In contrast, younger children—despite their improved sorting—typically could not express the basis of their sorting either verbally or with gestures. It is not unusual for young preschoolers to have trouble in explaining their behavior. Several studies have shown that labeling the relevant dimension increases its salience, and the same may be true of gesturing (Doebel & Zelazo, 2015). In the current study, the experimenter labeled the relevant dimension in both the Gesture and Control groups, but the Gesture group still sorted more accurately. Thus, it would seem that gesturing provides something beyond what a label provides. Delayed benefit of gesturing Although there were no age differences in the production of gestures overall, an interesting pattern in their benefit to younger children provides a clue to the process involved. Specifically, gesturing did not help the younger children until the second task, as shown in the ANOVAs, the regressions, and the differentiation of the Control and Gesture groups only after the switch in the trial-by-trial analysis. Importantly, this increase in sorting accuracy in young children from the first task to the second task was not accompanied by an increase in gesturing. This pattern of a delayed effect of gesturing resembles a pattern labeled a utilization deficiency observed in the strategy literature in which a strategy does not immediately help a child’s performance but does so eventually (Miller & Seier, 1994). In the strategy literature, a main interpretation of this phenomenon is that the early phase of acquiring a skill is very effortful and, thus, resource demanding. As a result, the new skill cannot at first be applied to improve performance. In the current study, as gesturing became less effortful for young children, they may have been more able to use it in the service of sorting. This delayed effect of gesturing is consistent with the theories described earlier proposing that gesturing sets in motion a process that benefits sorting. These generated processes take time. For example, the hierarchical competing systems model (Marcovitch & Zelazo, 2009) claims that experience with the task improves performance by encouraging reflection. Any reflection stimulated by gesturing takes time and might not lead to improved sorting until the second task. An alternative explanation of this delayed impact of gesturing comes from a study with third and fourth graders (Brooks & Goldin-Meadow, 2016). Practicing relevant gestures did not immediately improve children’s performance on mathematical equivalence problems, but it did increase children’s readiness to learn from subsequent instruction about mathematical equivalence. Similarly, in the current study, practice with gesturing during the instructions and the first task may have increased children’s readiness to learn about switching sorting rules on the second task, even though there was no explicit new instruction. One issue in the pattern of results concerns the fact that, among the older children, the substantial difference in mean performance between conditions did not reach significance. The reason for this is

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

55

not clear; a larger sample or a slightly younger older sample that was not already performing well might move this outcome to statistical significance. Overt gesturing versus covert processes Our trial-by-trial examination of the relation between gesturing and correct sorting can clarify whether gesture helped sorting (a) on a trial-by-trial basis, with more correct gestures on a particular trial associated with more correct sorts on that trial, or (b) in a more general way across the session. Our data favor the latter process. Specifically, the patterns of gesture production in Fig. 5 did not correspond to the patterns of sorting accuracy in Fig. 4. That is, overall, the Gesture group sorted more accurately than the Control group, but more children gesturing on a particular trial did not necessarily lead to more correct sorts on that trial, and the Gesture group performed well even on trials in which few children gestured. This pattern held even on the second task, where overall the younger children benefitted from gesturing. Thus, the current results raise an important question about the underlying process by which gestures facilitate EF. If gesturing facilitates EF in a more general way, one possibility is that the induced gesturing of the relevant dimension generated or strengthened a preexisting verbal representation of that dimension, as suggested by the greater number of verbal explanations in the Gesture condition than in the Control condition. Consequently, at least some children might no longer have needed to gesture or just gestured occasionally. Alternatively, their representation might be a covert gesture-like (motor-based) one. Piaget argued that when certain actions are repeated, they become an internalized action scheme. In addition, just as private speech gradually becomes internalized as inner speech, and private speech appears mainly on difficult tasks (Vygotsky, 1978), children who tend to gesture may use a mixture of overt and covert gestures, relying on overt gestures mainly when the task is more difficult. Thus, gesture-for-self may help self-regulation and EF in much the same manner that speech-for-self does. As support for the internalization of gestures, Chu and Kita (2011) found that the superior performance of gesturing adults continued even on trials when they no longer gestured overtly. They proposed that as the adults became more adept at the task, the ‘‘spatial computation supported by gestures becomes internalized, and the gesture frequency decreases” (p. 102). The decreased gesturing over trials during three of the four sets of trials in our study (see Fig. 5) is consistent with this hypothesis. The general impact of gesturing may even have occurred very early in the session; gesturing during instructions or on the easy pre-switch trials (when correct sorts are thought to take little effort) may have consolidated the sorting rule for the children. Imitating the experimenter’s gestures may have played some role in triggering understanding. It is known that perceiving another’s gesture can activate the observer’s motor system, which in turn facilitates the observer’s understanding of the person’s goal (Ping, Goldin-Meadow, & Beilock, 2014). In addition, when instructors used gestures while teaching a math concept, third and fourth graders adopted these gestures and later used them to explain their own strategies for these problems (Cook & Goldin-Meadow, 2006). The possibility that overt gesturing may lead to covert gestural or verbal representations also may explain why, in the older children, the relations between gesturing and accurate sorting were modest and typically did not reach significance. That is, some of the older children, who generally were sorting well, may have been using covert representations rather than overt ones and, thus, would have been scored as not gesturing, which would have diminished the apparent relationship between gesturing and sorting for them. Alternative explanations and future research Showing an effect of training always leads to the following two questions. Is the effect due to the element of training thought to be critical—in this case gesturing? If so, which element (or elements) of the training (instructions to gesture) was critical? Regarding the first question, the possibility that gesturing might simply have increased overall attention to the task is unlikely because children in the Control group were told to think hard about the task, which also should increase attention to the task. Moreover, Novack, Goldin-Meadow, and Woodward (2015) found that learning how to operate a novel

56

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

toy following demonstrations with iconic gestures was greater than when following point gestures. This finding suggests that iconic gesturing does not simply focus a learner’s attention; it also conveys information about how to solve a problem. The possibility that the Gesture condition had its effect simply by requiring children to take additional time before responding is unlikely because the correlation showed no relation between response time and sorting accuracy. In addition, although it is possible that the gesturing manipulation had its effect by clarifying which dimension was relevant by providing labels for the relevant dimension, children in both conditions were told numerous times (both before sorting began and again at the start of each trial) which dimension was relevant. Even if gesturing the post-switch relevant dimension made that dimension more salient, and thus facilitated sorting, this advantage should have been offset by the fact that gesturing the pre-switch relevant dimension should have increased its salience as well, which would encourage perseveration on that dimension and hinder post-switch sorting (see Doebel & Zelazo, 2015, on the importance of dimensional salience on the DCCS). Another point to address is that the condition difference could be due in part to hindering the performance of children in the Control group (e.g., by giving them the extra task of thinking hard) rather than improving the performance of the Gesture group. To explore this possibility, we compared the post-shift performance of the Control group with that of similarly aged children on the standard DCCS from a meta-analysis (Doebel & Zelazo, 2015), which reported percentages of children who passed the post-shift trials (typically sorted at least 5 of 6 post-shift trials correctly). With this criterion, 32.7% of the children in our Control group passed the post-shift trials, which is similar to the percentage reported by Doebel and Zelazo (2015; see Fig. 3) for our mean age of 44 months on the standard DCCS task. However, children in the meta-analysis were tested on dimensions and stimuli that differed from those in the current study. Thus, a more appropriate comparison is with a previous study of DCCS (O’Neill & Miller, 2013) using the same stimuli for one of our tasks (size/orientation) and drawing primarily from some of the same preschools as for most of the children in the current study. With the same criterion, in this previous study 46.0% passed the size/orientation post-shift trials, whereas 37.5% of the current sample passed. Considering that the previous sample was 3 months older than the current sample, their performance seems comparable. Thus, it appears unlikely that the condition difference found in the current study was due to hindering the performance of the Control group. Still another possibility—that gesturing simply prevents children’s hands from perseveratively pointing to a tray according to the first rule after the switch—is unlikely because the Gesture group outperformed the Control group overall, not just on the trials in which they gestured. Finally, it could be that needing to both gesture and sort facilitated performance by providing experience with responding twice on each trial—first implicitly in gesture and then by placing the card. However, it could be argued that the Control group also was given two tasks—to both think hard and sort. Still, in the Control group only sorting necessarily involved decision making, whereas both steps required decision making in the Gesture group. Future research addressing these concerns would help to clarify the role of gesture and ensure that an extraneous variable was not accounting for the differences observed between conditions. As for the second question, regarding the specific aspect of gesturing that is important, in future research it would be fruitful to examine separately several aspects of our gesture training—telling children to ‘‘think with their hands,” providing a gesturing model (the experimenter), and asking children to imitate/produce the gestures—to determine which elements are essential. At least for 6-year-olds, performing gesture is more beneficial, for problem solving, than observing gesture (Goldin-Meadow et al., 2012). Another feature that may matter is the timing of gesture training. For example, is it equally effective to have gesture practice before and during the sorting task? Regarding the positive relation between number of gestures and number of correct sorts among younger children on the second task in the Gesture condition, one possibility to be ruled out is that the children who are able to both gesture and sort correctly are already more advanced in EF. A DCCS pretest in future research could address this issue. Finally, future research could clarify whether gestures help performance only on tasks with dimensions that are easy to gesture such as shape, distance, shape, and orientation. Color, for example, would be difficult to gesture.

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

57

Conclusions The current study has important implications for research on both EF and gesture as well as for embodied cognition approaches more generally. For work on EF, the results open up a new avenue of research by suggesting that gestures are somehow involved in children’s EF—either as a contributor to the development of EF or as an aid to their performance on EF tasks. For example, gestures may strengthen or reorganize representations of information about a task in working memory. Or, gestures may facilitate performance by taking on some of the cognitive load during problem solving, thereby freeing cognitive resources to inhibit a prepotent response. For work on gestures, the results suggest an additional role for gestures—contributing to cognitive control. In addition, research on gestures might benefit from including a fine-grained trial-by-trial analysis. This analysis turned out to be illuminating in the current study, revealing patterns of change in gesturing and sorting accuracy that otherwise would have remained hidden and providing a clearer picture of the specific contributions of gesturing and age. Acknowledgment This article is based on Candace L. Rhoads’ master’s thesis. We thank Children’s Campus San Francisco State University, Montessori Children’s Center, Kids Konnect Child Development Center, and Sandbox Early Childhood Learning Center for participating in this study. References Alibali, M. W., Spencer, R. C., Knox, L., & Kita, S. (2011). Spontaneous gestures influence strategy choices in problem solving. Psychological Science, 22, 1138–1144. Bates, E., Camaioni, L., & Volterra, V. (1975). The acquisition of performatives prior to speech. Merrill-Palmer Quarterly, 21, 205–226. Benitez, V. L., Vales, C., Hanania, R., & Smith, L. B. (2017). Sustained selective attention predicts flexible switching in preschoolers. Journal of Experimental Child Psychology, 156, 29–42. Best, J. R., Miller, P. H., & Jones, L. L. (2009). Executive functions after age 5: Changes and correlates. Developmental Review, 29, 180–200. Boncoddo, R., Dixon, R., & Kelley, E. (2010). The emergence of a novel representation from action: Evidence from preschoolers. Developmental Science, 13, 370–377. Broaders, S. C., Cook, S., Mitchell, Z., & Goldin-Meadow, S. (2007). Making children gesture brings out implicit knowledge and leads to learning. Journal of Experimental Psychology: General, 136, 539–550. Brooks, N., & Goldin-Meadow, S. (2016). Moving to learn: How guiding the hands can set the stage for learning. Cognitive Science, 40, 1831–1849. Cameron, H., & Xu, X. (2011). Representational gesture, pointing gesture, and memory recall of preschool children. Journal of Nonverbal Behavior, 35, 155–171. Chu, M., & Kita, S. (2011). The nature of gestures’ beneficial role in spatial problem solving. Journal of Experimental Psychology: General, 140, 102–116. Cook, S. W., & Goldin-Meadow, S. (2006). The role of gesture in learning: Do children use their hands to change their minds? Journal of Cognition and Development, 7, 211–232. Davis, C. L., Tomporowski, P. D., McDowell, J. E., Austin, B. P., Yanasak, N. E., Naglieri, J. A., & Miller, P. H. (2011). Exercise improves executive function and academic achievement and alters neural activation in overweight children: A randomized controlled trial. Health Psychology, 30, 91–98. Deater-Deckard, K. (2014). Family matters: Intergenerational and interpersonal process of executive function and attentive behavior. Psychological Science, 23, 230–236. DeMarie-Dreblow, D., & Miller, P. H. (1988). The development of children’s strategies for selective attention: Evidence for a transitional period. Child Development, 59, 1504–1513. Doebel, S., & Zelazo, P. D. (2015). A meta-analysis of the Dimensional Change Card Sort: Implications for developmental theories and the measurement of executive function in children. Developmental Review, 38, 241–268. Ehrlich, S. B., Levine, S. C., & Goldin-Meadow, S. (2006). The importance of gesture in children’s spatial reasoning. Developmental Psychology, 42, 1259–1268. Göksun, T., Hirsh-Pasek, K., & Golinkoff, R. M. (2010). How do preschoolers express cause in gesture and speech? Cognitive Development, 25, 56–68. Goldin-Meadow, S. (2015). From action to abstraction: Gesture as a mechanism of change. Developmental Review, 38, 167–184. Goldin-Meadow, S., & Alibali, M. W. (2013). Gesture’s role in learning and development. In P. D. Zelazo (Ed.). The Oxford handbook of developmental psychology (Vol. 1, pp. 953–973). Oxford, UK: Oxford University Press. Goldin-Meadow, S., Cook, S. W., & Mitchell, Z. A. (2009). Gesturing gives children new ideas about math. Psychological Science, 20, 267–272. Goldin-Meadow, S., Levine, S. C., Zinchenko, E., Yip, T. K., Hemani, N., & Factor, L. (2012). Doing gesture promotes learning a mental transformation task better than seeing gesture. Developmental Science, 15, 876–884.

58

C.L. Rhoads et al. / Journal of Experimental Child Psychology 173 (2018) 41–58

Goldin-Meadow, S., Nusbaum, H., Kelly, S. D., & Wagner, S. (2001). Explaining math: Gesturing lightens the load. Psychological Science, 12, 516–522. Iverson, J. M., & Goldin-Meadow, S. (2001). The resilience of gesture in talk: Gesture in blind speakers and listeners. Developmental Science, 4, 416–422. Johnson, E. L., Munro, S. E., & Bunge, S. A. (2014). Development of neural networks supporting goal-directed behavior. In P. D. Zelazo & M. D. Sera (Eds.). Developing cognitive control processes: Mechanisms, implications, and interventions (Minnesota Symposia on Child Psychology (Vol. 37, pp. 23–54). New York: John Wiley. Kirkham, N., & Diamond, A. (2003). Sorting between theories of perseveration: Performance in conflict tasks requires memory, attention, and inhibition. Developmental Science, 6, 474–476. Kontra, C., Goldin-Meadow, S., & Beilock, S. L. (2012). Embodied learning across the life span. Topics in Cognitive Science, 4, 731–739. Krauss, R. M., Chen, Y., & Chawla, P. (1996). Nonverbal behavior and nonverbal communication: What do conversational hand gestures tell us? Advances in Experimental Social Psychology, 28, 389–450. Kuhn, L. J., Willoughby, M. T., Wilbourn, M. P., Vernon-Feagans, L., & Blair, C. B. (2014). Early communicative gestures prospectively predict language development and executive function in early childhood. Child Development, 85, 1898–1914. Marcovitch, S., & Zelazo, P. D. (2009). A hierarchical competing systems model of the emergence and early development of executive function. Developmental Science, 12, 1–25. Miller, P. H., & Seier, W. S. (1994). Strategy utilization deficiencies in children: When, where, and why. Advances in Child Development and Behavior, 25, 107–156. Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D. (2000). The unity and diversity of executive functions and their contributions to complex ‘‘frontal lobe” tasks: A latent variable analysis. Cognitive Psychology, 41, 49–100. Morsella, E., & Krauss, R. M. (2004). The role of gestures in spatial working memory and speech. The American Journal of Psychology, 117, 411–424. National Institutes of Health (2016). NIH Toolbox Retrieved from . Novack, M. A., Goldin-Meadow, S., & Woodward, A. L. (2015). Learning from gesture: How early does it happen? Cognition, 142, 138–147. O’Neill, G., & Miller, P. H. (2013). A show of hands: Relations between young children’s gesturing and executive function. Developmental Psychology, 49, 1517–1528. Pine, K. J., Lufkin, N., Kirk, E., & Messer, D. (2007). A microgenetic analysis of the relationship between speech and gesture in children: Evidence for semantic and temporal asynchrony. Language and Cognitive Processes, 22, 234–246. Ping, R., & Goldin-Meadow, S. (2010). Gesturing saves cognitive resources when talking about nonpresent objects. Cognitive Science, 34, 602–619. Ping, R. M., Goldin-Meadow, S., & Beilock, S. L. (2014). Understanding gesture: Is the listener’s motor system involved? Journal of Experimental Psychology, 143, 195–204. Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Cambridge, MA: Harvard University Press. Wagner, S. M., Nusbaum, H., & Goldin-Meadow, S. (2004). Probing the mental representation of gesture: Is handwaving spatial? Journal of Memory and Language, 50, 395–407. Wegner, D. M. (1989). White bears and other unwanted thoughts: Suppression, obsession, and the psychology of mental control. New York: Penguin. Wiebe, S. A., Morton, J. B., Buss, A. T., & Spencer, J. P. (2014). The emergent executive: A dynamic field theory of the development of executive function: II. A case study of the DCCS task. Monographs of the Society for Research in Child Development, 79(2), 12–25. Zelazo, P. D. (2006). The Dimensional Change Card Sort (DCCS): A method of assessing executive function in children. Nature Protocols, 1, 297–301.