Individual differences in executive function and learning: The role of knowledge type and conflict with prior knowledge

Abstract

Executive function (EF) predicts children’s academic achievement; however, less is known about the relation between EF and the actual learning process. The current study examined how aspects of the material to be learned—the type of information and the amount of conflict between the content to be learned and children’s prior knowledge—influence the relation between individual differences in EF and learning. Typically developing 4-year-olds (N = 61) completed a battery of EF tasks and several animal learning tasks that varied on the type of information being learned (factual vs. conceptual) and the amount of conflict with the learners’ prior knowledge (no prior knowledge vs. no conflicting prior knowledge vs. conflicting prior knowledge). Individual differences in EF predicted children’s overall learning, controlling for age, verbal IQ, and prior knowledge. Children’s working memory and cognitive flexibility skills predicted their conceptual learning, whereas children’s inhibitory control skills predicted their factual learning. In addition, individual differences in EF mattered more for children’s learning of information that conflicted with their prior knowledge. These findings suggest that there may be differential relations between EF and learning depending on whether factual or conceptual information is being taught and the degree of conceptual change that is required. A better understanding of these different relations serves as an essential foundation for future research designed to create more effective academic interventions to optimize children’s learning.

Introduction

A large body of research shows that executive function (EF)—higher-order neurocognitive skills such as working memory, cognitive flexibility, and inhibitory control—is related to and predicts school readiness and academic achievement (e.g., Allan, Hume, Allan, Farrington, & Lonigan, 2014; Best, Miller, & Naglieri, 2011; Blair & Razza, 2007; Jacob & Parkinson, 2015). In addition, intervention research shows that EF can be trained and improved in certain contexts (e.g., Diamond & Ling, 2016). This has motivated researchers and educators to target EF as a way to improve young children’s success in school and to help close achievement gaps. Despite this desire to use EF interventions to improve academic achievement, there is little support for a causal relation between EF and academic outcomes (Jacob & Parkinson, 2015) and more research is needed to determine how EF relates to learning. In particular, more research is needed to better understand how specific aspects of the information being learned influence the relation between individual differences in EF and learning. The goal of the current study was to begin to investigate the relation between EF and preschoolers’ learning by examining two characteristics of the content to be learned: (a) the type of knowledge and (b) the amount of conflict between the information to be learned and the learner’s prior knowledge. Understanding how these factors influence the relation between EF and learning would help contribute to research that can inform the creation of more effective academic interventions and instructional strategies for young children before beginning formal schooling.

Executive function

EF skills are higher-order cognitive skills used during goal-directed behaviors (Carlson, Zelazo, & Faja, 2013). The development of EF coincides with the development of the prefrontal cortex, with the preschool period being a time of rapid EF skill growth (Carlson, 2005; Casey, Giedd, & Thomas, 2000). Three aspects of EF have been emphasized in previous research: working memory, inhibition, and cognitive flexibility (Miyake et al., 2000). Although research with older children and adults has suggested that there are three main components of EF, there is mixed evidence for the structure of EF skills in younger children. For example, some research has suggested that EF is a unitary construct (e.g., Wiebe et al., 2011), whereas other research has found that preschoolers’ performance on EF tasks maps onto two or three factors (e.g., Espy, Kaufmann, McDiarmid, & Glisky, 1999; Hughes, 1998). The inconsistencies across studies have been attributed to factors such as the use of different EF tasks, the way in which EF is measured (e.g., direct assessment, questionnaire), and difficulty in finding tasks that measure only one component of EF (i.e., the task impurity problem) (Miller, Giesbrecht, Müller, McInerney, & Kerns, 2012). Despite a lack of consensus on the structure of EF during the preschool period, there is a wealth of research showing that EF predicts important outcomes such as academic achievement (Carlson et al., 2013).

EF and academic achievement

Most of the research on the role of EF in learning has focused on the relation between EF and standardized achievement measures in math, literacy, and science across development (e.g., Allan et al., 2014; Clements, Sarama, & Germeroth, 2016; Jacob & Parkinson, 2015; Raghubar, Barnes, & Hecht, 2010; Yeniad, Malda, Mesman, Van Ijzendoorn, & Pieper, 2013). The main finding from this body of literature is that higher EF skills are associated with and predictive of higher academic achievement even when controlling for IQ. During the preschool period, there is evidence that EF predicts academic achievement both concurrently and longitudinally. For example, a meta-analysis that included 75 studies with preschool and kindergarten children found that children’s inhibitory control skills were predictive of their concurrent math and literacy achievement scores (Allan et al., 2014). There is also evidence that early EF skills predict later academic achievement in math and literacy. For example, Bull, Espy, and Wiebe (2008) found that children’s EF measured in preschool predicted their math achievement at 7 years of age. For literacy, Alloway and Alloway (2010) found that working memory at 5 years of age was a better predictor of literacy achievement at 11 years than IQ.

When examining the relation between EF and academic achievement, some researchers have made the distinction between tasks that require the use of EF skills from the traditional cognitive perspective and those that require using EF skills to regulate one’s behaviors or motor responses, with both being important for academic success (e.g., McClelland et al., 2014). The latter tasks have been described as those that measure behavioral regulation. Researchers have argued that behavioral regulation skills are essential for certain classroom behaviors such as raising one’s hand to answer a teacher’s question instead of blurting out the answer and being able to sit still while learning (McClelland & Cameron, 2012). There is also evidence that children’s behavioral regulation skills are related to their literacy, vocabulary, and math skills concurrently and longitudinally in preschool and kindergarten (McClelland et al., 2007, 2014; Ponitz, McClelland, Matthews, & Morrison, 2009).

The majority of the literature on the relation between EF and academic skills has used standardized measures of academic achievement. However, academic achievement is only one way to measure students’ success in school (Neuenschwander, Rothlisberger, Cimeli, & Roebers, 2012). For example, grades and classroom behaviors that are optimal for learning also are indices of success. In addition, it is important to consider how academic achievement measured by standardized tests might be different from learning or constructing new knowledge (Modrek, Kuhn, Conway, & Arvidsson, 2019; Neuenschwander et al., 2012). For example, to score high on an academic achievement test, students need to express knowledge they have already acquired after the learning process has occurred. Therefore, it is important to consider other ways in which EF could influence learning and success in school.

Potential roles that EF plays in learning

There are several different roles that EF might play in learning. One possibility is that EF influences learning indirectly by allowing students to engage in behaviors that are optimal for learning. Learning-related behaviors increase opportunities for children to be engaged in instructional activities and include participating, being able to work together successfully with teachers and peers, refraining from behavior that would disrupt learning activities, and paying attention (Fantuzzo, Perry, & McDermott, 2004; Nesbitt, Farran, & Fuhs, 2015; Razza, Martin, & Brooks-Gunn, 2015). Research shows that children’s EF is related to these learning-related behaviors (Brock, Rimm-Kaufman, Nathanson, & Grimm, 2009; Neuenschwander, Röthlisberger, Cimeli, & Roebers, 2012; Rimm-Kaufman, Curby, Grimm, Nathanson, & Brock, 2009) and that children’s learning-related behaviors mediate the association between their EF and academic achievement (e.g., Baptista, Osório, Martins, Verissimo, & Martins, 2016; Nesbitt, Farran, & Fuhs, 2015; Sasser, Bierman, & Heinrichs, 2015). The idea that EF allows learners to be ready to learn and to be compliant in the classroom is often discussed in professional development and policy-related materials as an important way in which EF helps with success in school (e.g., Ackerman & Friedman-Krauss, 2017).

Another potential role that EF plays in learning is that EF skills are needed for children to “show what they know.” For example, if students who have learned how to add and subtract are completing a math worksheet that has a mix of addition and subtraction problems, they will need to be able to flexibly switch between them to accurately express their knowledge of both operations. Another way in which EF might influence the expression of knowledge is by dictating the types of strategies children use. For example, there have been studies showing that children select different strategies depending on their EF while solving arithmetic problems (Barrouillet & Lépine, 2005) and that children with higher EF are more likely to choose the more optimal arithmetic strategy than children with lower EF (Lemaire & Lecacheur, 2011).

At an even deeper level, EF skills might be important for processing new information and constructing new knowledge. For instance, studies have shown that EF, especially inhibitory control, predicts children’s ability to use newly learned strategies that are more effective than previously learned strategies when solving algebraic and arithmetic problems (Khng & Lee, 2009; Robinson & Dubé, 2013). There is also evidence showing that individual differences in EF moderate the effectiveness of an intervention for gains in academic knowledge, with children with higher EF benefitting the most (Bascandziev, Powell, Harris, & Carey, 2016; Bascandziev, Tardiff, Zaitchik, & Carey, 2018; Kolkman, Hoijtink, Kroesbergen, & Leseman, 2013; Laski & Dulaney, 2015; Miller, Rittle-Johnson, Loehr, & Fyfe, 2016; Rhodes et al., 2014, 2016). EF might also be important for determining which instructional strategies or approaches are most helpful when individuals are constructing new knowledge. For example, a study with high school students found that students with lower EF showed better transfer of learning from a chemistry computer simulation that was designed with a guided instruction approach, whereas students with higher EF exhibited better transfer when the simulation had an exploratory learning approach (Homer & Plass, 2014). Although there are several different potential roles that EF might play in children’s success in school, the current study focused on those related to the learning process itself such as constructing new knowledge during the preschool period.

EF and construction of knowledge

There is previous research showing that EF plays a role in the construction of new conceptual knowledge in science (Bascandziev et al., 2016, 2018) and math (Kolkman et al., 2013; Laski & Dulaney, 2015; Miller et al., 2016), wherein children with higher EF learn more in these domains, even when controlling for IQ or verbal ability. Moreover, some studies suggest that there might be different relations between EF and learning depending on whether the task requires children to express knowledge or to construct knowledge, with evidence showing that working memory might be especially important for constructing new knowledge (Bascandziev et al., 2016; Miller et al., 2016). This research is informative given that most of the correlational studies examining the relation between EF and learning used standardized academic achievement tests, which appear to align more with children’s ability to express their existing knowledge. Therefore, the current study examined preschoolers’ construction of new knowledge about animals to add to our understanding of the role that EF plays in the learning process. This is important to better understand how learning interventions could be altered to be more beneficial for all students.

Type of knowledge being learned

Although there is a growing body of evidence showing that individual differences in EF predict children’s expression and construction of knowledge, the majority of this research has focused on learning conceptual information. This may be due in part to the fact that researchers began to explore whether EF was a domain-general skill that could be used to explain how conceptual change occurs (Bascandziev et al., 2018; Carlson & Moses, 2001; Carlson, Claxton, & Moses, 2015; Tardiff, Bascandziev, Carey, & Zaitchik, 2020; Vosniadou, 2014; Zaitchik, Solomon, Tardiff, & Bascandziev, 2016). Most conceptual change researchers posit that individuals start with a naïve theory that must be restructured to form more scientifically accurate theories (Carey, 2009; Shtulman & Valcarcel, 2012; Vosniadou, 2014; Wellman & Gelman, 1992). Children’s naïve theories are formed based on their experiences in the world. Starting formal instruction in math and science presents them with information that conflicts with their naïve theories. When this occurs, children need to restructure their conceptual knowledge and categories to allow their initial theories to coexist with newer theories (Carey, 2009; Vosniadou, 2014). Some have proposed that children use Quinian bootstrapping to add knowledge to their original theories, change connections between different types of representations, and ultimately change their concepts (Carey, 2009; Zaitchik et al., 2016). According to this idea, EF might allow children to recognize conflicts between new knowledge and their own theories, inhibit old theories to use more accurate theories, and switch back and forth between different theories depending on the situation (Vosniadou, 2014; Vosniadou et al., 2015; Zaitchik et al., 2016).

Recently, researchers have begun to extend the study of the domain-general processes involved in conceptual change by making the distinction between different types of knowledge acquisition; some types of information might be easier to learn than others, thereby relying on different cognitive skills. For example, the distinction between “knowledge enrichment” and “conceptual construction” has been used by some researchers (Bascandziev et al., 2018; Tardiff et al., 2020). Knowledge enrichment refers to instances where the information to be learned can be understood using concepts and words the learner already has, whereas conceptual construction refers to instances where the learner does not yet have the concepts or words to understand the content being learned. Evidence from a few studies looking at knowledge construction has suggested that EF is more strongly related to learning conceptual information than to learning factual information (Bascandziev et al., 2018; Rhodes et al., 2014, 2016; Tardiff et al., 2020). For example, a vitalist biology intervention for early school-aged children was more beneficial for children with higher EF than for those with lower EF when learning conceptual knowledge. However, individual differences in EF were not related to improvements in learning facts about animals. Instead, children’s verbal ability was found to be related to their learning of factual information (Bascandziev et al., 2018). Moreover, a recent study investigated the expression of conceptual knowledge related to vitalist biology with 5- and 6-year-olds and included a standardized measure of factual knowledge. They found that EF was related to children’s vitalist biology conceptual knowledge even after controlling for their factual knowledge scores (Tardiff et al., 2020).

There also have been a few training studies that included teaching and measuring adolescents’ learning of both factual and conceptual knowledge in the domains of biology and chemistry (Rhodes et al., 2014, 2016). Rhodes et al. (2014) found that students’ factual learning about DNA was significantly correlated with their performance on a planning EF task (Tower of London). Their conceptual learning was significantly correlated with their performance on the planning task and a spatial working memory task (Rhodes et al., 2014). Contrary to the findings from Bascandziev et al. (2018), the authors found that students’ EF was related to both their conceptual learning and factual learning after the intervention, but they did not control for verbal ability or IQ (Rhodes et al., 2014). Another training study by Rhodes et al. (2016) investigated the relation between 12- and 13-year-olds’ EF and factual and conceptual learning about chemistry. The researchers found a significant correlation between the students’ conceptual learning and their performance on EF tasks that measured planning and spatial working memory, controlling for age and verbal ability. In this case, consistent with the study by Bascandziev et al. (2018), children’s EF was not a significant predictor of their learning of factual information (controlling for age and verbal ability). However, Rhodes et al. (2016) did not account for children’s prior knowledge, so it is unclear whether they were measuring gains in knowledge due to the intervention or capturing differences in preexisting knowledge.

These studies suggest that the relation between individual differences in EF and learning might depend on whether children are asked to learn factual knowledge or conceptual knowledge, with a stronger relation for the latter. However, more research is needed to see whether these findings replicate given that the majority focused on vitalist biology specifically and other studies did not control for IQ (Rhodes et al., 2014) or students’ prior knowledge related to the intervention (Rhodes et al., 2014, 2016). Therefore, the current study included both factual and conceptual learning tasks and controlled for prior knowledge and verbal IQ to better understand how the role of EF might differ depending on the type of information children are expected to learn.

Conflict with prior knowledge

Lastly, we considered whether overcoming conflicts with prior knowledge could be a reason why EF skills are related to conceptual change. Theoretically, information processing skills are essential for progressing through the stages of conceptual development. EF allows individuals to engage in the type of information processing that is necessary to hold two ways of thinking about something in mind and to think about them in conflict with each other. There are at least three different relations between prior knowledge and new information that can occur during learning (Chi, 2008). One possibility is that learners have no prior knowledge and the material being learned is completely novel. It is also possible that learners have some prior knowledge related to the material to be learned, but the new information does not conflict with their prior knowledge and instead adds to it. A third possibility is that the new information conflicts with learners’ prior knowledge, and they must overcome or revise their prior knowledge for learning to occur. These three common conditions under which children learn new information provide a means to further explore the role that EF plays in learning given that they might rely on EF skills to different degrees.

Previous research has focused on two of these possibilities: no conflict with prior knowledge (knowledge enrichment) and conflict with prior knowledge (conceptual change or construction). However, studies have not included tasks in which children have no prior knowledge. Therefore, the current study created learning tasks that had a range of conflict with the learner’s prior knowledge (no prior knowledge, no conflicting prior knowledge, and conflicting prior knowledge) to see whether the role of EF in learning increased as the amount of conflict in the task increased. By manipulating how much conflict is present in the learning tasks, we were able to examine how increasing the information processing demands of the task by requiring children to overcome prior knowledge influences the relation between EF and learning. To more fully understand how individual differences in EF might be contributing to learning, we aimed to investigate aspects of the learning process that have not been studied extensively in previous research, such as characteristics of the content to be learned, and to do so during a critically important period of development just prior to entering formal schooling.

The current study

In this study, 61 4-year-olds completed a series of EF, learning, and IQ tasks. A within-person design was used such that all children received all the learning tasks to examine how individual differences in EF related to their learning of information about animals that differed in our two main variables of interest: (a) type of knowledge and (b) amount of conflict with prior knowledge. The current study had three aims to better understand the relation between individual differences in young children’s EF and learning. The first was to determine whether individual differences in children’s EF predicted their overall learning performance. We hypothesized that children’s EF skills would be a significant predictor of their overall learning, with children, with higher EF skills having higher learning scores. The second aim was to examine how individual differences in EF predicted children’s factual learning versus conceptual learning. Based on previous studies that measured both factual and conceptual learning in older children and adolescents (Bascandziev et al., 2018; Rhodes et al., 2014, 2016; Tardiff et al., 2020), we predicted that individual differences in EF would predict conceptual learning but not factual learning. The third aim of the current study was to investigate whether the amount of conflict between the information to be learned and the learner’s prior knowledge would interact with individual differences in EF to predict learning. To study this aim, we created three conceptual learning tasks that theoretically differed in the amount of conflict with the learner’s prior knowledge: no prior knowledge, prior knowledge but not conflicting, and conflicting prior knowledge. Based on previous research (Bascandziev et al., 2018; Tardiff et al., 2020) and theoretical connections between conceptual change and EF, we hypothesized that EF would be a stronger predictor of children’s conceptual learning when there was conflict with the learner’s prior knowledge. If individual differences in EF have different relations to learning depending on the type of information being learned and the extent to which it conflicts with prior knowledge, this would suggest that there may be certain learning contexts in which is it most useful to target children’s EF to improve academic outcomes.

Method

Participants

Families were recruited from a university database in a midwestern metropolitan region of the United States. Institutional review board approval was obtained by the same university. Participants included 61 typically developing 4-year-olds (29 girls; M_age = 53.72 months, SD = 2.71, range = 49–59). The sample was mostly Caucasian (77%), but some participants identified as Asian (1.6%), African American (1.6%), White Hispanic (11.5%), and biracial (8.2%). The majority of the sample was also upper-middle class (median annual family income = $125,000–$149,000), with 3.3% earning $50,000 or less, 24.6% earning $50,000-$100,000, 37.7% earning $100,000-$150,000, and 34.4% earning more than $150,000 annually. The sample was also well educated, with 49.2% of primary caregivers having a graduate or professional degree, 3.3% having some graduate school experience, 34.4% having a bachelor’s degree, 9.8% having some college experience, and 3.3% having a high school diploma. An additional 5 children participated but were excluded due to prematurity (n = 1), speech and motor delays (n = 1), not enough proficiency with English (n = 1), or noncompliance/refusal to complete all the learning tasks (n = 2).

Procedure

Children participated in the lab individually for one 75-min videotaped session. Parents completed a demographics questionnaire. Children completed an animal naming task, a series of learning tasks about animals (counterbalanced), three EF tasks, and a verbal IQ measure. Tasks were administered by a trained female graduate student. Parents received a $10 gift card and children received a T-shirt and a small toy for participating.

Measures

Animal naming task (5 min)

To measure children’s familiarity with animals, children were shown 17 pictures of animals that varied in how familiar they are to children and adults. For example, a low-difficulty animal would be a pig, a medium-difficulty animal would be an ostrich, and a high-difficulty animal would be an axolotl. The experimenter presented each picture and asked children to name each animal. They were told they might not have seen some of the animals before, so they could say, “I don’t know.” Children received 1 point for each correctly labeled animal (0–17). Videos were double-coded, and reliability was acceptable (intraclass correlation coefficient [ICC] = .87).

Learning tasks

Children completed four learning tasks, which are described below (see also Table 1).

Table 1.

Schematic showing differences across the different factual and conceptual learning tasks.

Knowledge type	Amount of conflict	Topic	Description of learning Task	How learning was measured
Factual	No prior knowledge	Novel animal facts	Children are told a story about pangolins with pictures. The story included facts about pangolins such as where they live, what they eat, and what they do when they are scared.	Children are asked comprehension questions (e.g., “Where do pangolins live?”) and application questions (i.e., shown four pictures of different items, such as a mouse, ladybug, cricket, and grass, and asked to identify all the things pangolins would eat).
Conceptual	No prior knowledge	Fantastical animal categories	Children learn about fantastical animals that do not exist in the real world and how to identify them based on their characteristics. For example, they are shown a fantastical animal called a “mufi” and are told they could tell it is a “mufi” because it has long legs.	Children are shown pictures of fantastical animals for three different fantastical animal groups. For each picture, children are asked whether it is the target animal (e.g., “mufi”) or not and why.
	No conflicting prior knowledge	Mammals	Children are taught that some animals are part of the mammal family because they have characteristics of mammals (e.g., warm blooded, have hair/fur, have bone in their back). They are introduced to an animal expert called Mr. Hippo (a hippo puppet), and the experimenter demonstrates asking Mr. Hippo questions to help determine whether an animal from a picture is a mammal or not. For example, looking at a picture of an elephant, the experimenter would ask Mr. Hippo, “Do elephants have warm blood?”	Children are shown pictures of animals and are asked to identify whether each is a mammal or not and why. They also have the option to ask Mr. Hippo any questions to help them determine whether an animal is a mammal or not
	Conflicting prior knowledge	Biological inheritance	Children are taught that animals’ identities are based on their insides and who their parents are instead of their outside appearance by giving examples of animals that look like one animal (e.g., a pig) but have the insides of another animal and parents that are another animal (e.g., a bear).	Children are told stories about animals that look like one animal on the outside but have the insides and parents that are another animal. Children are then asked whether they think the animal is really whatever animal it looks like on the outside and why they think that.

Note. The corresponding author can be contacted for more information about the specific stimuli used in these tasks.

No prior knowledge, factual learning task (10 min).

Children were told a story with pictures about pangolins (novel animal) that included facts such as where they live, what they like to eat, what they do when they are scared, and whether they are nocturnal. A pangolin was included in the animal naming task to see whether children were familiar with and could name the pangolin. After the story, children were asked two memory check questions to ensure that they were paying attention (e.g., “What was the pangolin’s name in the story?”), four comprehension questions (e.g., “What do pangolins do when they are scared?”), and two application questions (e.g., “Point to all the things that pangolins would like to eat”). Children received a memory check score out of 2, a comprehension score out of 4, and an application score out of 2. Inter-rater reliability was excellent for each of the three scores (ICCs = .90–.98).

No prior knowledge, conceptual learning task (10 min).

Children were taught about a new animal group (fantastical animals created in Piekny & Maehler, 2013) and features that make an animal part of this new animal category (e.g., red fur, wings). Children were first given a practice block of trials in which they were taught how to play the game. Here they were shown a picture of an animal and were told it was called a “mufi” because it had long legs. Children were then shown three more pictures of animals one at a time and were asked whether they were a mufi or not and why. The animals varied in whether they had long legs but also had other characteristics of the first picture shown (e.g., fur). Then children were given three blocks of three test trials, with each block consisting of a new fantastical animal with a defining characteristic. Children received 1 point for every correct answer on test trials (0–9). Their reasons also were coded, and children received 1 point if they gave a correct reason for why an animal was or was not the target novel animal (0–9). Therefore, children received an accuracy score and a reasoning score. Inter-rater reliability was excellent (ICCs = .94–.97).

No conflicting prior knowledge, conceptual learning task (10 min).

Children were taught about the category of mammals and the features that can be used to categorize an animal as a mammal (e.g., warm blooded, has hair/fur, has bone in its back). They were then shown two different animals (one mammal and one nonmammal) and were told characteristics about the animal that are needed to determine whether it is a mammal. Then children were asked whether the animal was a mammal or not and why. During these two practice trials, they were given feedback. Next, children were introduced to a puppet described as an animal expert named Mr. Hippo. They were then told they would be shown pictures of animals and that their job was to decide whether each one was a mammal or not. If they were not sure, they were told they could ask Mr. Hippo any questions to help them figure it out. The experimenter then demonstrated asking Mr. Hippo a question to determine whether an elephant was a mammal or not (e.g., “Mr. Hippo, do elephants have warm blood?”) and invited children to ask Mr. Hippo a question. After the experimenter and/or children asked three questions (warm-blooded, bone in back, and hair/fur), children were asked whether they thought an elephant was a mammal or not and why (with corrective feedback). During the test trials, children were shown seven pictures of animals individually and were asked whether they thought each was a mammal or not and why. Any questions they asked Mr. Hippo were recorded. Lastly, children were asked how they know whether an animal is a mammal. Children’s responses were coded for accuracy (0–7). Children’s reasons for why an animal was a mammal or not also were coded, and children received 1 point if they correctly reasoned that an animal did or did not have all three of the characteristics they were taught (0–7). The number of questions children asked Mr. Hippo was also coded. Inter-rater reliability was excellent for each score (ICCs = .94–.99).

Conflicting prior knowledge, conceptual learning task (10 min).

Children were taught about instances of familiar animals that were in conflict with their prior knowledge. We adapted stimuli from Keil’s (1992) study of children’s concepts of inheritance and insides/essentialism. Children were told stories about animals that look like one kind of animal on the outside (e.g., raccoon) but, after being studied by scientists, it was discovered that they had the insides (e.g., blood, bones, brain) of another animal (e.g., skunk) and their moms and dads were a different type of animal (e.g., skunk). Children were then asked whether the animal was really a [skunk] or a [raccoon] and why. They were given one pretest trial to assess whether they already understood the concepts of essentialism and inheritance. Children were then shown another example and were taught that insides are important regardless of what the animal looks like on the outside and that the parents’ identities are important for determining what kind of animal it is (learning trial). Two test trials followed. For each, children were asked what they thought the animal really was and why they thought that. Children’s responses were coded, and children received a pretest score (0–1), a learning trial score (0–1), and a test trial score (0–2). In addition, children’s reasons on test trials were coded, and children received a score of 1 if they mentioned the animals’ insides or parents/babies or a score of 0 if they gave no response or another reason (0–2). Inter-rater reliability was excellent for each of the scores (ICCs = .98–1.00).

Executive function

Although there is considerable overlap among the different components in EF, as noted earlier, we included a battery of measures to represent each aspect of EF (cognitive flexibility, working memory, and inhibition).

Minnesota Executive Function Scale (5 min).

The Minnesota Executive Function Scale task (MEFS; Carlson & Zelazo, 2014) was included as a measure of cognitive flexibility. The MEFS is administered on an iPad and has seven levels of difficulty. The task is reliable and valid and can be used with children as young as 2 years. This version was normed on a U.S. sample of 32,800 typically developing children aged 2 to 18 years (Carlson, 2020). Children were asked to sort virtual cards based on different rules or dimensions (e.g., sort by color, sort by shape), with some levels requiring children to switch rapidly between rules. Children’s starting level was determined by their age, and then children either went up or down in levels depending on their performance until a basal and ceiling were reached. The MEFS software algorithm calculates a total raw score (0–100) based on children’s accuracy and response times.

Backward Word Span (5 min).

The Backward Word Span task (BWS; Carlson, Moses, & Breton, 2002) measures children’s working memory. Children were asked to repeat a list of words read out loud by the experimenter in reverse order. Before test trials, children were given one practice trial and were given feedback if they answered incorrectly. Children were given up to four attempts to get the practice trials correct (and then proceeded regardless). On test trials, lists began with two words and then levels increased by one word if children responded correctly. Children had three chances to correctly repeat lists at each level. Children received a score corresponding to the highest number of words correctly recalled, with possible scores ranging from 1 to 7.

Statue task (2 min).

The Statue task (Korkman, Kirk, & Kemp, 1998), a subtest of the NEPSY (A Developmental NEuroPSYchological Assessment) standardized battery, assesses children’s ability to stay still and to inhibit impulses and motor responses. It was included as a measure of inhibitory control. Children were asked to stand like a statue holding a pretend flag in one hand at a 90-degree angle with their eyes closed for 75 s. Meanwhile, the experimenter produced sounds and distractions (e.g., coughing, dropping a pencil on the table) at specified times. Body movements, vocalizations, and eye openings were coded as errors during 5-s intervals. Involuntary coughing, silent smiling, small movements of the fingers, and subtle movements due to balance issues were not coded as errors. For each interval, children received a 2 (no errors), 1 (one error), or 0 (two or more errors). If children said they were done pretending to be a statue and wanted to quit the task, they received 0 for the remaining intervals. Scores could range from 0 to 30.

Stanford–Binet Intelligence Scales for Early Childhood (10 min)

We used the verbal knowledge subtest of the Stanford–Binet Intelligence Scales for Early Childhood–Fifth Edition (Roid, 2005) to control for children’s verbal IQ in the analyses. The starting level for 4-year-olds consisted of showing them different toys and figurines (e.g., ball, cat) and asking them to label them. Then children were shown pictures in which people are doing different actions (e.g., drinking, running) and were asked to verbally state what the person was doing. Finally, they were asked to define words, and definitions were scored on a scale of 0 to 2 using test manual procedures until they reached a ceiling of four consecutive 0 scores. Standard scores with a mean of 10 and a standard deviation of 3 were used in analyses.

Results

Missing data

We had missing data for two of our target variables: the Stanford–Binet Verbal standard scores (3.28% missing) and the Statue task scores (9.84% missing). The latter was due mainly to children declining to try the task or not wanting to close their eyes during the task, rendering their data invalid. Given the Stanford–Binet Verbal subtest was the last task in the study procedure, missing data were due to children declining to participate, or not having sufficient time in the session to complete the task. Data appeared to be missing completely at random (MCAR) based on Little’s (1988) MCAR test, χ²(20) = 13.93, p = .83. Analyzing raw data is considered an acceptable method of handling missing data when the data are MCAR (Peng, Harwell, Liou, & Ehman, 2006). Furthermore, the results did not differ when we conducted our main analyses using imputed data (with multiple imputation). Therefore, we present the results analyzing the raw data below for ease of interpretation.

Preliminary results

As shown in Table 2, there was sufficient range on each measure of interest, with no floor or ceiling effects. Correlations for the raw data are shown in Table 3. We first examined intercorrelations among the three EF measures. BWS and MEFS were significantly correlated (r = .42). However, scores on the Statue task did not significantly correlate with the other measures. Therefore, we created a working memory + cognitive flexibility (WM + CF) composite by averaging the standardized BWS and MEFS scores. Children’s Statue task scores were kept separately as an inhibitory control (IC) score. In addition, an overall learning composite score was created by standardizing and then averaging children’s scores on all four learning tasks. An overall conceptual learning composite score was created by averaging children’s scores on the three conceptual learning tasks: no prior knowledge, no conflicting prior knowledge, and conflicting prior knowledge.

Table 2.

Descriptive statistics for all measures.

Measure	n	Range	M (SD)
Age in months	61	49–59	53.72 (2.71)
Stanford-Binet Verbal IQ (raw score)	59	5–15	10.08 (2.59)
BWS (highest level)	61	1–4	1.82 (0.85)
MEFS (max = 100)	61	17–78	46.22 (10.34)
Statue (max = 30)	55	0–30	21.11 (8.47)
Animal familiarity task (max = 17)	61	3–12	7.90 (1.53)
No conflict, factual: Memory check questions (max = 2)	61	0–2	1.34 (0.70)
No conflict, factual: Comprehension questions (max = 4)	61	0–4	3.15 (1.03)
No conflict, factual: Application questions (max = 2)	61	0–2	0.82 (0.72)
No prior knowledge, conceptual: Practice trials score (max = 3)	61	0–3	1.43 (0.78)
No prior knowledge, conceptual: Test trials total score (max = 9)	61	5–9	7.31 (1.21)
No prior knowledge, conceptual: Test trials reasoning score (max = 9)	61	0–9	4.52 (2.62)
No conflicting prior knowledge, conceptual: Pretest (max = 1)	61	0–1	0.05 (0.22)
No conflicting prior knowledge, conceptual: Practice trials score (max = 3)	61	0–3	1.79 (0.71)
No conflicting prior knowledge, conceptual: Test trials total score (max = 8)	61	0–7	4.13 (1.49)
No conflicting prior knowledge, conceptual: Test trials reasoning score (max = 8)	61	0–7	2.21 (2.25)
No conflicting prior knowledge, conceptual: Number of questions asked Mr. Hippo (max = unlimited)	60	0–5	0.47 (0.98)
Conflicting prior knowledge, conceptual: Pretest score (max = 1)	61	0–1	0.21 (0.41)
Conflicting prior knowledge, conceptual: Learning trial score (max = 1)	61	0–1	0.26 (0.44)
Conflicting prior knowledge, conceptual: Posttest/test total score (max = 2)	61	0–2	1.05 (0.87)
Conflicting prior knowledge, conceptual: Reasoning score for test trials (max = 2)	61	0–2	0.48 (0.81)

Note. MEFS = Minnesota Executive Function Scale. BWS = Backward Word Span.

Table 3.

Bivariate correlations for study variables.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. Child age	1
2. Child gender (female)	.06	1
3. SB Verbal IQ (N = 59)	−.10	.01	1
4. Prior knowledge	.25*	−.12	.16	1
5. BWS	.14	−.07	.20	.12	1
6. MEFS	.15	.16	.12	.27*	.42**	1
7. Statue task (N = 55)	−.14	.33*	.17	−.02	.00	.19	1
8. WM + CF composite	.17	.05	.19	.23	.84**	.84**	.12	1
9. Factual learning total score	.08	.09	.35**	.25	.12	.27*	.36**	.23	1
10. Conceptual, no prior knowledge: Total score	.06	.16	.55**	.33**	.25	.30*	.35**	.32*	.43**	1
11. Conceptual, no conflicting prior knowledge: Total score	.07	.16	.19	.53**	.20	.36**	.06	.33**	.18	.45**	1
12. Conceptual, conflicting prior knowledge: Total score	.05	.11	.29*	.54**	.29*	.39**	.14	.40**	.18	.57**	.51**	1
13. Overall conceptual learning composite	.07	.17	.42**	.57**	.30*	.42**	.22	.43**	.32*	.82**	.80**	.84**	1
14. Overall learning composite	.09	.17	.47**	.56**	.29*	.45**	.31*	.44**	.61**	.83**	.73**	.77**	.95**	1

Note. N = 61 unless otherwise noted. SB Verbal IQ = Stanford–Binet Intelligence Scale, Verbal subtest. BWS = Backward Word Span. MEFS = Minnesota Executive Function Scale. WM + CF = working memory + cognitive flexibility.

^*p < .05.

^**p < .01.

Given that the main goal of the current study was to measure children’s learning of new information, we controlled for children’s prior knowledge. Although it was not possible to have prior knowledge about the fantastical animals in the conceptual, no prior knowledge task (because they do not exist in the real world), a prior knowledge score was created by averaging scores on the other three learning tasks that taught children about information that they might have had prior knowledge about before the study session (e.g., knowing about pangolins or the concept of mammals and inheritance). This prior knowledge score was used as a covariate in all analyses. Without controlling for children’s prior knowledge, it would be challenging to interpret whether children’s EF skills were predictive of their learning or whether children with higher EF already had sufficient prior knowledge.

Preliminary analyses also indicated that there were no significant effects of gender or learning task order on children’s learning, so these variables were not included in the following models.

Main analyses

To address our aims, we conducted a series of hierarchical linear regressions to determine whether individual differences in EF predict children’s learning after controlling for variables that have been found to be related to children’s EF. This allowed us to test the unique contribution of children’s EF for their learning above and beyond the covariates such as prior knowledge, verbal IQ, and age.

Overall learning

To address our first aim of whether individual differences in EF relate to children’s overall learning of new information, we created an overall learning composite score. This score was created by standardizing and averaging children’s total scores (accuracy plus reasoning scores) on each of the four learning tasks. We used hierarchical linear regressions predicting overall learning composite scores. As shown in Table 4, children’s WM + CF composite scores were a significant predictor of children’s overall learning composite scores, controlling for age in months, verbal IQ, prior knowledge, and IC score (B = .23, p = .01). Children’s IC scores also uniquely predicted overall learning composite scores, controlling for age in months, verbal IQ, prior knowledge, and WM + CF composite scores (B = .02, p = .03). Children’s age in months and verbal IQ accounted for 25% of the variance in children’s overall learning. When children’s prior knowledge was added to the model, an additional 21% of the variation was explained. Finally, in Block 3, children’s WM + CF composite and IC scores accounted for an additional 13% of the variance in children’s overall learning composite scores, controlling for age in months, verbal IQ, and prior knowledge (Table 4).

Table 4.

Regression analyses with EF predicting overall learning composite score.

Variable	B (SE)	t	p	R 2	ΔR2
Block 1				.25	–
Age in months	.05 (.03)	1.48	.15
SB Verbal IQ	.14 (.04)	4.01	<.001
Block 2				.46	.21***
Age in months	.01 (.03)	0.21	.84
SB Verbal IQ	.10 (.03)	3.19	.002
Prior knowledge score	1.57 (0.36)	4.38	<.001
Block 3				.59	.13**
Age in months	.01 (.03)	0.19	.85
SB Verbal IQ	.08 (.03)	2.77	.008
Prior knowledge score	1.42 (0.33)	4.34	<.001
WM + CF composite	.23 (.09)	2.66	.01
IC score	.02 (.01)	2.27	.03

Note. N = 54. EF =executive function. SB Verbal IQ = Stanford–Binet Intelligence Scale, Verbal subtest. WM + CF = working memory + cognitive flexibility. IC = inhibitory control.

^**p < .01.

^***p < .001.

Type of knowledge

We next examined whether individual differences in EF predicted learning conceptual information versus factual information. To address this second aim, we created an overall conceptual learning composite by averaging the standardized scores on each of the three conceptual tasks. We then used a series of hierarchical linear regressions predicting children’s overall conceptual learning and factual learning separately. For each model, we included age in months and verbal IQ in Block 1, prior knowledge in Block 2, and the two EF measures in Block 3. In the model predicting children’s overall conceptual learning, we found that only the WM + CF composite scores were a significant predictor when controlling for age in months, verbal IQ, and IC scores (B = .26, p = .01). The covariates (age in months, verbal IQ, and prior knowledge) accounted for 45% of the variance in children’s overall conceptual learning. Adding WM + CF and IC scores to the models explained an additional 9% of the variance (Table 5). We next examined the model predicting children’s factual learning and found that performance on the inhibitory control task was a significant predictor when controlling for age in months, verbal IQ, and WM + CF composite scores (B = .06, p = .01). The covariates (age in months, verbal IQ, and prior knowledge) accounted for 16% of the variance in children’s factual learning. Adding WM + CF and IC scores to the model explained an additional 13% of the variance in children’s factual learning (Table 6).

Table 5.

Regression analyses with EF predicting overall conceptual composite score.

Variable	B (SE)	t	p	R 2	ΔR2
Block 1				.20	–
Age in months	.05 (.04)	1.20	.24
SB Verbal IQ	.14 (.04)	3.47	.001
Block 2				.45	.25***
Age in months	−.01 (.03)	−0.18	.86
SB Verbal IQ	.09 (.04)	2.56	.01
Prior knowledge score	1.92 (0.41)	4.72	<.001
Block 3				.54	.09*
Age in months	−.01 (.03)	−0.33	.74
SB Verbal IQ	.07 (.03)	2.14	.04
Prior knowledge score	1.73 (0.39)	4.48	<.001
WM + CF composite	.26 (.10)	2.57	.01
IC score	.01 (.01)	1.28	.21

Note. N = 54. EF = executive function. SB Verbal IQ = Stanford–Binet Intelligence Scale, Verbal subtest. WM + CF = working memory + cognitive flexibility. IC = inhibitory control.

^*p < .05.

^***p < .001.

Table 6.

Regression analyses with EF predicting factual learning total score.

Variable	B (SE)	t	p	R 2	ΔR2
Block 1				.15	–
Age in months	.09 (.07)	1.23	.22
SB Verbal IQ	.22 (.08)	2.83	.01
Block 2				.16	.01
Age in months	.06 (.07)	0.87	.39
SB Verbal IQ	.19 (.08)	2.44	.02
Prior knowledge score	.80 (.89)	0.91	.37
Block 3				.29	.13*
Age in months	.08 (.07)	1.09	.28
SB Verbal IQ	.15 (.08)	2.00	.05
Prior knowledge score	.74 (.85)	0.86	.39
WM + CF composite	.20 (.23)	0.89	.38
IC score	.06 (.02)	2.57	.01

Note. N = 54. EF = executive function. SB Verbal IQ = Stanford–Binet Intelligence Scale, Verbal subtest. WM + CF = working memory + cognitive flexibility. IC = inhibitory control.

^*p < .05.

Amount of conflict with prior knowledge

To address our final aim of investigating whether individual differences in EF interact with the amount of conflict there is between children’s prior knowledge and information to be learned, we used hierarchical linear modeling to account for children’s within-person variation in their performance on the three conceptual learning tasks that varied in the amount of conflict with children’s prior knowledge: no prior knowledge, no conflicting prior knowledge, and conflicting prior knowledge. We ran two analyses: one looking at the interaction between task conflict and WM + CF composite scores and the other looking at the interaction between task conflict and IC scores. We found a significant interaction between WM + CF composite scores and the amount of conflict in the learning task, controlling for children’s age in months, verbal IQ, prior knowledge, and IC scores (see Table 7). As illustrated in Fig. 1, children with lower EF performed worse compared with children with higher EF on the conflicting prior knowledge conceptual task (B = .12, p = .003), whereas the difference in learning performance between children with lower and higher EF for the other two tasks was not as prominent. Post hoc simple slopes analyses revealed that only the slope for the conflicting prior knowledge task was significantly different from zero (B = .15, p ≤ .001). In contrast, we did not find a significant interaction between task conflict and IC scores, controlling for children’s age in months, verbal IQ, prior knowledge, and WM + CF composite scores.

Table 7.

Hierarchical linear model analyses to examine the interaction between working memory + cognitive flexibility composite scores and task conflict for children’s learning scores.

Variable	B (SE)	t	p
WM + CF composite	.03 (.03)	0.87	.38
IC score	.003 (.002)	1.16	.25
No conflicting prior knowledge	−.27 (.04)	−7.68	<.001
Conflicting prior knowledge	−.26 (.04)	−7.22	<.001
Age in months	−.004 (.01)	−0.46	.64
SB Verbal IQ	.02 (.01)	1.96	.06
Prior knowledge score	.47 (.10)	4.91	<.001
WM + CF Composite × No Conflicting Prior Knowledge	.004 (.04)	0.09	.93
WM + CF Composite × Conflicting Prior Knowledge	.12 (.04)	2.95	.003

Note. N = 54. EF = executive function. WM + CF = working memory + cognitive flexibility. IC = inhibitory control. SB Verbal IQ = Stanford–Binet Intelligence Scale, Verbal subtest.

Fig. 1.

Interaction between working memory + cognitive flexibility (WM + CF) composite scores and the amount of conflict with prior knowledge in the learning task, controlling for age in months, verbal IQ, prior knowledge, and inhibitory control (IC) scores (N = 54).

Discussion

The current study further expands our knowledge of the relation between individual differences in EF and young children’s learning by examining different aspects of knowledge that children are asked to learn. By focusing on the type of knowledge (factual vs. conceptual) and the amount of conflict present between children’s prior knowledge and the information to be learned, the current study replicated and extended previous findings on the relation between EF and learning.

As we predicted, individual differences in all EF measures used predicted children’s learning across tasks (factual and conceptual) even after controlling for age in months, verbal IQ, and children’s prior knowledge. This finding is significant because it demonstrates that individual differences in EF uniquely predict children’s overall learning after accounting for other variables that have been found to be related to both EF and learning. This finding adds to previous literature on the relation between individual differences in EF and children’s construction of knowledge instead of focusing on children’s ability to express knowledge they have already acquired.

The second main finding was that different types of EF measures predicted children’s learning of different types of information. Individual differences in working memory and cognitive flexibility predicted children’s conceptual learning, whereas their performance on an inhibitory control task with relatively low demands on working memory and cognitive flexibility (Statue task) predicted their learning of factual information. This was contrary to our hypothesis based on previous research, which found that verbal ability or measures of IQ, but not EF, were predictive of learning factual information (Bascandziev et al., 2018; Tardiff et al., 2020). Those authors interpreted their findings by stating that learning conceptual information involves overcoming conflicts, which theoretically would require EF skills, whereas there is no conflict in learning facts about a certain domain. One reason our findings were different might be due to the types of EF tasks used in the current study compared with those used in previous studies. The Statue task used in the current study is a measure of motor inhibition that might serve as a proxy for being able to sit still while learning, thereby representing an indirect effect of EF on learning. If this is the case, it is possible that performance on the Statue task predicted learning of factual information because learning facts requires that children sit still and pay attention. However, being able to sit still might not be enough to learn conceptual information because, in that case, children must rely on their working memory and cognitive flexibility skills to overcome previous conflicting knowledge to learn about a new concept.

It is also possible that children’s performance on the Statue task may have been influenced by factors other than their ability to inhibit their motor responses. For example, children’s anxiety or fearfulness of being in a new environment (i.e., the lab) with a person they just met (i.e., the experimenter) might have influenced their willingness to engage in the task and to follow certain rules of the task such as keeping their eyes closed. The Statue task also might have captured children’s ability to comply with the experimenter’s directions during a task where it is more obvious when they have not followed directions than the other EF tasks used (MEFS and BWS). During those tasks, the experimenter does not remind children that they need to follow the rules during the test trials; however, in the Statue task protocol, the experimenter gives children verbal reminders that they are supposed to be acting like a statue if they make any errors. Previous studies that found EF only predicted learning of conceptual information used shifting, inhibition, and working memory tasks that did not require motor inhibition and were similar to the tasks included in our WM + CF composite (Bascandziev et al., 2018; Tardiff et al., 2020). Therefore, more research is needed to determine why more complex cognitive EF tasks (e.g., MEFS and BWS used in this study) would predict learning conceptual information, whereas a simple motor inhibition task (i.e., Statue task) would predict learning of factual information. Moreover, the differences among EF components and tasks that require cognitive versus behavioral regulation need to be further examined in future research to better understand the relations between EF and learning of factual information versus conceptual information and the findings from the current study.

The third main finding of the current study is that the amount of conflict between learners’ prior knowledge and the information to be learned significantly interacted with individual differences in EF to predict children’s conceptual learning. Using a continuum of conflict (no prior knowledge and no conflict, prior knowledge but no conflict, and prior knowledge and conflict), we were able to sort out the conditions under which EF plays a role in learning in relation to the amount of prior knowledge children have and whether the information to be learned conflicts with any prior knowledge. Our findings extend previous research by suggesting that the conflict between learners’ prior knowledge and the information to be learned might be crucial for EF to play a role in children’s construction of new knowledge. More specifically, interaction effects showed that working memory and cognitive flexibility skills were especially predictive of children’s performance on the conflicting prior knowledge conceptual task, such that children with higher WM + CF composite scores tended to have higher learning scores. Therefore, the amount of conflict with one’s prior knowledge does seem to be one reason why there is a relation between EF and learning of conceptual information.

Limitations and future directions

In addition to its strengths, the current study had a number of limitations. One limitation is that the study included only 4-year-olds. This was intended to focus on the year prior to formal school entry so as to assess their “readiness” to learn and to study a narrow age range so as to maximize variability in learning (constraining floor and ceiling effects), but we were unable to look at developmental change and how relations between EF and learning might change with age. For example, as children develop, different aspects of EF may be more important for their learning of different types of information. Prior research with older children (5- and 6-year-olds) found that EF predicted their learning of conceptual information but not factual information (Bascandziev et al., 2018; Tardiff et al., 2020). However, in our study, we found that motor inhibition (Statue task) predicted young children’s learning of factual information. Given that younger children are still developing their motor inhibition skills, children’s ability to sit still while learning might be more relevant for learning during the preschool period. In contrast, as children get older and acquire more knowledge, there could be more interference from prior knowledge to be overcome in the process of revising and constructing new theories, wherein working memory and cognitive flexibility skills are most relevant. Future research should include longitudinal studies that can examine how the relation between EF and learning changes over time.

Another limitation was that there might have been some conflict inherent in conditions that we theoretically thought would have no conflict with children’s prior knowledge. For example, during the mammals task (conceptual, no conflicting prior knowledge task), some children expressed conflict with their prior knowledge when they were resistant to having an animal also be called a mammal (e.g., a bear could not be a mammal because it is a bear). This task was designed to teach children that certain animals could also be part of a group called mammals (i.e., enrichment of prior knowledge) rather than to challenge their existing concepts, but it also might have challenged their emerging understanding of class inclusion (e.g., Markman & Callanan, 1983). Despite anecdotally observing children struggling with this type of conflict, there was still a good range of performance on the task (no floor or ceiling effects) such that the task seemed appropriate for this age group. It is also important to note that children’s individual differences in their WM + CF composite scores predicted their learning only on the conceptual, conflicting prior knowledge task and not on the conceptual, no conflicting prior knowledge task when we looked at the individual conceptual learning tasks as the dependent variables. Therefore, there might be different types of conflict with one’s prior knowledge where some may be harder to overcome than others and, in turn, rely more on EF skills. Future research is needed to explore the different types and amount of conflict with learners’ prior knowledge using different types of learning tasks and experimental methods to determine more precisely how individual differences in EF are related to learning.

In addition, the current study included a sample of children mostly from high-income families. Given previous research showing that EF may be especially important for children growing up in high-risk contexts (Razza, Martin, & Brooks-Gunn, 2010; Wright, Masten, & Narayan, 2013) and that EF mediates the association between socioeconomic status and academic achievement (Lawson & Farah, 2017), it is important to conduct future research to see whether the findings replicate with more diverse samples.

Our study also was limited in that we developed learning tasks that could be completed and post-tested in one 75-min visit to the lab. Future research should explore how the relations between EF and learning might differ when looking at retention of learning by having participants return to the lab at a later date and assessing their knowledge of the topics taught or by conducting the study in a more ecologically valid classroom setting.

Finally, although the current study illustrated the importance of EF for different types of learning, it was correlational. It provides a foundation for future experimental training and intervention studies to more directly connect the dots between EF and learning.

Conclusion

Overall, we found evidence that individual differences in EF predict 4-year-olds’ learning of facts and concepts about animals above and beyond children’s age in months, verbal IQ, and prior knowledge. We also found that children’s working memory and cognitive flexibility skills are important for their learning of conceptual information, whereas their motor inhibition was related to their ability to learn factual knowledge. Lastly, our study suggested that the amount of conflict with learners’ prior knowledge interacts with children’s working memory and cognitive flexibility skills to predict learning such that higher EF is more beneficial during tasks where there is more conflicting old information to be reflected on and reconciled with new information. These findings provide an essential spring-board for future research to better understand how aspects of the learning environment, such as characteristics of the learning goals (e.g., type of knowledge, amount of conflict with prior knowledge), interact with individual differences in EF to better understand the relation between EF and learning. A better understanding of how EF influences learning will allow researchers to design more effective academic and EF interventions to improve young children’s academic outcomes.