Universality and Context-Specificity in Early Executive Function Development

Abstract

A renewed interest in early executive function (i.e., EF or the conscious control of thought and behavior) development has led several research groups to suggest that EF may be emerging and is less coordinated (e.g., showing few relations between tasks) in the first few years (Devine et al., 2019; Gago Galvagno et al., 2021; Johansson et al., 2016; Miller & Marcovitch, 2015; Ribner et al., 2022). This potentially universal development in EF does not exclude the possibility that EF may also differ across context (e.g., Gago Galvagno et al., 2021; Lohndorf et al., 2019; Tran et al., 2015) reflecting unique strengths and development built within one’s sociocultural environment. The present paper explores potential universal and context-specific early EF developments by focusing on three aims: 1) reviewing work on EF within the first two years of life that may speak to potential universality in the measurement, structure, growth, stability, and conceptualization of early EF 2) reviewing research that may speak to how the sociocultural context may play a role in context-specific development within early EF and 3) examining potential developmental EF frameworks for understanding universal and context-specific developments of early EF within context.

Executive function (i.e., EF or the conscious control of thought and behavior) is a widely researched topic with work focusing on development increasing fivefold during the first decade of the 21^st century (Carlson et al., 2013). Most developmental studies focus on the rapid development exhibited within the preschool years (e.g., Blair & Razza, 2007; Carlson, 2005; Garon et al., 2008; Isquith et al., 2005; Wolfe & Bell, 2004; Zelazo et al., 1997, 2003) showing improvements in tasks assessing working memory (i.e., WM or the ability to hold and manipulate increasing amounts of task relevant information in mind over longer delays that can be used to guide behavior), inhibition (i.e., suppressing prepotent or affectively driven behaviors), and shifting abilities (i.e., flexibly switching responses and attention between task relevant information; see Garon et al., 2008; Jacques & Marcovitch, 2010 for reviews) that all contribute to problem-solving. Despite the plethora of work, there are several unanswered questions and areas that have been noted as important to consider as we move forward in the study of EF development.

For one, we have fewer studies about EF development in the first few years. In 2022, a search of “executive function” in PsycInfo revealed nearly 6.5 times more articles for the preschool years compared to children from 2-23 months. Although the preschool years are considered an important period for understanding early developmental shifts in EF, there are indications that children younger than 2 years demonstrate controlled, goal-directed behavior. By 6 months, children begin to search for objects hidden in one of two locations (Pelphrey & Reznick, 2002), improve in the amount of information they can hold in mind (Pelphrey et al., 2004), and the length of the delay they can tolerate before search (Diamond, 1985; Diamond & Doar, 1989). Later in the first year, children succeed on a more difficult search task, the A-not-B task, where children must shift search response to a new location (B) after finding a toy hidden in location A multiple times (Diamond, 1985; Marcovitch & Zelazo, 1999; Piaget, 1954). There are also several other cognitive and social achievements emerging during this period that likely require controlled behavior (see Wiebe et al., 2010 for a discussion of related behaviors), although not always studied as “EF”. For instance, during the first few years of life children regulate emotions (Mangelsdorf et al., 1995; Nigg, 2017), delay gratification (e.g., Kochanska et al., 1998), imitate complex sequences (e.g., Alp, 1994; Wiebe & Bauer, 2005), begin to demonstrate means-end behavior (e.g., Cheng et al., 2018), and control motor behavior and action (e.g., Adolph et al., 2009). Thus, more work with this age range is warranted.

Another limitation to understanding EF development is that we have fewer studies addressing basic methodological and conceptual issues in early EF. This may partially be due to the single-task methodology used for assessing EF in a young sample with limited linguistic and attentional abilities. It has been suggested that approaching the study of EF with a battery of tasks when using behavioral assessments is an important method for addressing the task impurity problem (i.e., no pure EF task exists given the involvement of other task demands like linguistic knowledge, see Miyake & Friedman, 2012; Wiebe et al., 2008). Assessing EF with multiple tasks alleviates concern regarding impurity for any single task and allows for latent factor approaches where researchers can explore whether an EF factor or factors emerge from performance across multiple tasks. A battery of tasks also allows researchers to address foundational questions likely related to universal elements of EF, like EF’s functional organization or structure. For instance, Miyake and Friedman (2012) suggested that adults’ performance on any EF tasks involves what they term “common EF” related to the ability to form and maintain task relevant representations that guide lower-level processes in service of a goal. This common EF overlaps with the inhibitory component (as inhibition may be byproduct of the ability to execute goal-related processing rather than an independent process, Friedman & Miyake, 2017) with additional abilities required in shifting (i.e., ability to select, apply, and update the correct tasks goal) or updating (i.e., ability to precisely update relevant information in working memory), depending on the task. The structure of EF has also been examined in children, though there are suggestions that it looks different compared to adults (Wiebe et al., 2008, 2011). Comparatively, fewer studies have examined EF with multiple measures in a sample 2 years of age or younger consistent with a latent factor approach. Diamond et al. (1997) conducted one of the first studies examining cognitive control using multiple measures at 15, 18, and 21 months as part of a longitudinal study on the cognitive functioning of children treated for Phenylketonuria (PKU) and matched controls through 7 years of age. This seminal work found that performance on a modified A-not-B task and 3-boxes tasks (a WM task where children had to find 3 identical toys in 3 distinct boxes) improved across the second year, though tasks were not well correlated with each other.

Finally, several researchers have suggested that it is necessary for future studies of EF to be better rooted in context. For instance, Doebel (2020) suggested that EF may be better conceptualized as skills related to control in service of a goal—with the key being that goals are influenced by other contextual factors like prior knowledge, the individual’s beliefs, norms, and preferences (see also Sarma & Thomas, 2020). Raver and Blair (2020) suggest the field needs to focus on studying EF while simultaneously considering the varied contexts in which EF operates. They suggest this may include shifting focus from a deficit model (e.g., considering the costs of sociocultural factors like poverty and stress to development) to one that considers how stressors may be linked to structural forms of inequality. Further, they suggest that prioritizing culture and community in research is important, and one should approach families and communities as agents with their own strengths when working toward interventions and solutions. Although research related to EF in the first few years has moved forward in the study of EF and the contextual factors that surround it, more work is needed.

In the present paper, we consider these potential areas of future research. Given that less is known about the emergence and development of early EF, we focus our paper on the first two years of life from birth to approximately 2 years of age. We first review studies that have measured EF via a battery of tasks so that we can examine foundational (and potentially universal) questions surrounding how best to conceptualize EF in the first couple years. In addition, we review the potential for context-specific development related to considering how EF may be influenced by sociocultural factors. Finally, we explore how current developmental EF frameworks may account for potential universal and context-specific developments of early EF within context.

1.0. Search Strategy

In the present paper, we conducted a scoping review recommended for work with aims of better understanding the volume of work, methodology, synthesis, and gaps in the literature (Bruce & Bell, 2022; Munn et al., 2018). We chose this approach because although EF appears to be examined less in the first few years of life, the volume of work is unclear—especially for studies utilizing a behavioral battery of tasks across varied context.

Articles were included for review if they (1) included at least 2 behavioral assessments of EF in their methods (to allow for examination of relations between measures), (2) had a group of participants with an average age of approximately 2-years or younger (we included studies with an average age of up to 26 months), (3) were published in English, (4) were published in a peer-reviewed journal (5) contained empirical data (i.e., were not a review, commentary), and (6) were published after Diamond et al.’s 1997 seminal monograph (though the search did not reveal any articles published before Diamond et al., 1997). Studies examined typical development during the first two years. The article search was conducted from November 12^th 2022 through November 14^th 2022 across the electronic database PsycInfo specific to the discipline of psychology. Because our question was specific to a battery of EF tasks, we used the search term “executive function” and advanced search options in the database to apply equivalent subjects (to broaden our search) and selected the infancy age group in the advanced search (2-23 months, though this search yielded articles up to 26 months which we considered. This search resulted in 41 studies summarized in Table 1.

Table 1.

Studies with children 26 months of age and younger employing and battery of EF tasks

Authors (Year)	Demographics	Participants & Design	Tasks	Structure
1) Earliest assessments at approximately 2-years (24 +/− 3 months) of age
Bernier, Carlson, Bordeleau, & Carrier (2010)	Canadian Metropolitan Area 86.7% Caucasian 68.3% college degree Average $70,000 income (Canadian)	60 infants (36 girls) M=26.3 months, SD=0.8 Longitudinal EF at 2 timepoints (18 month only had 1 measure)	26 Months *Spin the pots *Delay of Gratification *Shape Stroop *Baby Stroop	* 2-factor solution with PCA *Factor 1: impulse control-DoG *Factor 2: conflict EF- spin pots, baby Stroop, shape Stroop
Bernier et al., (2014)	Canadian Metropolitan Area 87.3% Caucasian 61.9% college degree, $20,000 to $100,000 income (Canadian)	63 infants (27 girls) M=25.4 months, SD = 1.1 Longitudinal EF at 1 timepoint	25 Months *Spin the pots *Delay of Gratification *Shape Stroop *Baby Stroop	*2-factor solution with PCA *Factor 1: impulse control-DoG *Factor 2: conflict EF- spin pots, shape Stroop, baby Stroop *Conflict & Impulse control r=.28**
Bernier et al., (2012)1	Canadian Metropolitan Area 87% mothers and 79% fathers Caucasian 58.1% mothers, 56.5% fathers college degree Mean income roughly between $60,000-$79,000 (Canadian)	62 infants (38 girls) M=26.1 months Longitudinal EF at 2 timepoints (2- and 3-years)	2-years *Spin the pots *Delay of gratification *Shape Stroop *Baby Stroop	* 2-factor solution with PCA *Factor 1: impulse control-DoG *Factor 2: conflict EF- spin pots, baby Stroop, shape Stroop
Braren et al., (2021)2	United States (N=131) United Kingdom (N=221) Netherlands (N=132) 89.7% mothers and 91.7% fathers White 5.36 mothers and 5.09 fathers average education (out of 7 max) 2.70 (couple’s average income proportional to median income)	484 infants (49.3% girls) M=24.46 months, SD=.81 Longitudinal EF at 1 timepoint	24 Months *Multilocation search task *Ball run task *Baby Stroop task	*Latent variable framework (CFA with Bayesian estimator) to get bayes estimates for each of 3 EF Tasks. *multistep search was significant correlated with ball run, r=.24**, and baby Stroop, r=.31.
Broomell & Bell (2022)	Mid-Atlantic Region of the United States 77.6% White 61.7% college degree or higher	410 infants (209 girls) (305 infants at 24-month visit) Longitudinal EF at 5, 10, 24, 36, 48 months and 9 years (5 &10 months only used 1 task)	24-Months *A-not-B *Simon Says *DCCS *Tongue task *Crayon Delay	*A-not-B related to DCCS pre-switch, r=.15* *DCCS pre-switch related to Simon says, r=−.18*, tongue task r=.17**, and crayon task, r=.16** *Simon says related to the tongue task, r=−.29** *Tongue task related to the crayon task, r=.13* *4 correlations were not significant *Manifest variables loaded onto EF factor in SEM
Carlson et al., (2004).	Metropolitan Seattle Area, United States Predominantly White Modal education was a college degree	81 children (41 girls) M=24 months, SD = 1.41 Longitudinal EF at 2 and 3 years	24-Months *Reverse categorization *Multilocation Search *Shape Stroop *Snack Delay *Gift Delay	*Shape Stroop related to reverse categorization, r=.33*, and snack delay r=.40** *Snack delay related to gift delay r=.43** *2 correlations were marginally significant *5 correlations were not significant *An EF composite was created based on PCA
Cheng et al., (2018)	Beijing, China Average education at least college degree Median income between $36,000 and $72,000 Chinese Renminbi per year	96 infants (54 females) M=25.1 months, SD=1 Longitudinal EF at 25 and 38 months	25-Months *Spin the pots *reverse categorization *Externally imposed delay	*none of the tasks were related, rs<.10 *individuals scores for each task considered separately
Cuevas & Bell (2014)3	United States Original Sample (n=202) 85.3% Caucasian 72.3% Mothers and 65.6% fathers completed a bachelor’s degree or higher (of those reporting)	152 infants M=2.1 years/ 25.2 months, SD=22 days Longitudinal EF at 2 years, 3 years, and 4 years	25-Months *A-not-B with invisible displacement *Tongue Task	*the first principal component for the 2 tasks explained 55% of variance *Individual indicators were standardized, averaged and standardized again to yield a EF composite Z score.
Gago Galvagno et al., (2019) 4	Autonomous City and Province of Buenos Aires, Argentina 100% Argentine 58.33% Unsatisfied Basic Needs (low SES)	60 infants (34 girls) 18 to 24 Months EF Measured at 1 time point	18- to 24 Months *A-not-B with multiple locations *Spatial reversal *Snack Delay	*Perseverative A-not-B and spatial reversals correlated, r=.53. *No other EF task performance correlated *Individual scores for each task considered separately
Gago Galvagno et al., (2021) 4	Autonomous City and Province of Buenos Aires, Argentina 91.33% Argentine 61% Unsatisfied Basic Needs (low SES)	80 infants (42 girls) M=21.33 months, SD=2.70 Cross-sectional, EF Measured at 21 months and 33 months	21 Months *A-not-B with multiple locations *Spatial reversal task *Snack Delay	*More errors on A-not-B related to more perseveration on Spatial Reversal, r=.52** *No other EF task performance was correlated *individual scores for each task considered along with number of EF tasks passed
Joyce et al., (2016)	USA 74% mothers and 68% fathers college degree 90% Caucasian	81 participants (38 girls) M=2.10 years, SD=.05 Longitudinal at 2, 3, and 4 years	2-years *Tongue task *Crayon delay *Simon Says *Dimensional Change Card Sort (DCCS)	* Collapsed into composite scores *Conflict composite (Simon Says, DCCS) *Delay Inhibitory Control (Tongue task, Crayon task) *Conflict and delay were not correlated
Kochanska et al., (2000)	United States (original sample n=112) Families were mostly Caucasian and represented a range of socioeconomic status. 59% mothers and 57% fathers completed college or postgraduate work, 58% earned more than $40,000/year	106 infants (52 girls) M=22.30 months, SD=.55 Longitudinal EF at 22 and 33 months	22 Months *Snack Delay *Wrapped Gift *Gift in a Bag *Walk-a-Line-Slowly *Tower *Shapes Task	* Cronbach’s alpha between tasks was modest .42, average item total correlation was .27. *Individual task performance considered
McHarg et al., (2020)2	East of England, United Kingdom 42% mothers and 39% fathers had more than a bachelor degree Subjective social status mothers=3.67 and fathers=4.33 (1=worst to 10=best)	179 participants at T1 (79 girls) M=24.29months, SD=.85 Longitudinal EF at 24 and 36 months	24-Months *Multilocation Search Task *Ball Run *Baby Stroop	*EF score created by summing together the number of tasks each child passed 24-Months *reliability coefficient alpha=.49 (based on tetrachoric correlations)
Neale et al., (2018)	Cambridge, United Kingdom 89% White British Ethnicity or Nationality	36 participants (18 girls) Visits at 12, 18, and 24 months. Longitudinal EF with multiple measures at 24 months	24-Months *Gift Delay *Snack Delay	*Gift delay and snack delay converted to z score and merged to composite.
Pauen & Bechtel-Kuehne (2016)	Southwest Germany Middle-class	87 children (gender equally distributed) 22 months (n=45) 24 months (n-42)	22- and 24-Months *Shape and Color Sort *Forbidden Cookie *Find the Toy	*individuals scores for each task considered separately *No significant correlations between individual components found
Wolfe et al., (2014)3	Large Rural University in mid-Appalachian region, United States Original Sample (n=211, 112 girls, 89% Caucasian) 72% mothers and 65% fathers had college degrees.	153 infants at 2 years Longitudinal EF with multiple measures at 2 years	2-years *DCCS (preswitch only) *Crayon Task	*Composite z-score score was created *The first principal component explained 64% of the variance at 2 years
Zhang et al., (2022)5	Beijing, China 85% mothers and 78% fathers had bachelor’s degree or higher	163 infants (82 girls) M=755days, SD=33 Longitudinal EF measured at 1 timepoint	24-Months *Multilocation search *Shape Stroop Task *Reverse Categorization Task *Delay of Gratification Task	24-Months *individuals scores for each task considered separately *No significant correlations between individual components found
2) Earliest assessment at approximately 1.55 years (18 +/− 3 months)
Bernier, Carlson, & Whipple (2010)1	Canadian Metropolitan Area 81.3% Caucasian 78.8% college degree Primarily Middle Class, Average $70,000 (Canadian)	80 infants (44 girls) M=18.3 months M=26.3 months Longitudinal EF at 2 timepoints (18 months only had 2 measures)	18-Months *Hide the pots *Categorization 26-Months *Spin the pots *Delay of gratification *Shape Stroop *Baby Stroop	18-Months *Two EF tasks at 18 months were unrelated, r=.11, considered separately 26-Months *task scores mildly to moderately correlated * 2-factor solution with PCA *Factor 1: impulse control-DoG *Factor 2: conflict EF- spin pots, baby Stroop, shape Stroop *No significant correlations between 2 EF dimensions
Frick et al., (2018)	Uppsala, Sweden Initial sample (n=124) 75.4% mothers and 63.4% fathers had college or university degree	117 at 18 months M=18.06, SD=26 EF measured at 1 time point	18 Months *Prohibition task *A-not-B Task	*Prohibition was not related to A-not-B at 18 months, Spearman’s Rho = .01
Gago Galvagno, De Grandis et al., (2022)4	Autonomous City and Province of Buenos Aires, Argentina 100% Argentine Low- and Mid-SES	75 infants M=20.97 months, SD=2.40 EF Measured at 1 time point	21 Months *A-not-B with multiple locations *Spatial reversal *Snack Delay	*Perseverative A-not-B was correlated with perseveration on Spatial Reversal, r=.47** *No other task EF performance was correlated *Individual scores for each task considered along with number of EF tasks passed
Gago Galvagno, Miller, et al., (2022)4	Autonomous City and Province of Buenos Aires, Argentina 100% Argentine 55.7% Unsatisfied Basic Needs (Low-SES)	70 infants (37 girls) M=20.97, SD=2.40 EF Measured at 1 time point	21 Months *A-not-B with multiple locations *Spatial reversal task *Snack Delay	*Perseverative A-not-B correlated with perseveration on spatial reversal, r=.60** *Other correlations not reported *Individual scores for each task considered along with number of EF tasks passed
Marciszko et al., (2020)6	Uppsala, Sweden 62% mothers and 52% fathers university degree	118 children (50% girls) M=544 days, SD=12 Longitudinal, EF measured at 1 timepoint	18 months *Prohibition task *Tricky box *Hide and seek	*The three EF tasks were uncorrelated *Individual scores for each task considered
Poulin-Dubois et al., (2022)	Large Metropolitan city, Canada 60 monolingual, 42 bilingual 59% reported household income $75,000 or higher 73% Bachelors degree or higher	102 neurotypical infants (47 girls) M=17.25 months, SD=.98 EF measured at 1 time point	17-months *Detour reaching *Delayed response *Multiple location	*Individual scores for each task considered *Relationship between tasks not considered
Tu et al., (2022) 6	Uppsala, Sweden 62% mothers and 52% second parent university degree Most participants lived in White middle-class families living in a University town	104 infants (53 boys) M=544 days, SD=12.1 Longitudinal EF measured at 18 and 30 months	18-Months *Prohibition task *Tricky Box *Hide and seek *Delayed gratification *Reversed Categorization	*Individual scores for each task considered * There were no significant correlations between scores of tasks within and between age groups
3) Earliest assessment at approximately one year (12 months +/− 3 months)
Blum & Ribner (2022)2	New York City, United States Middle- to high- income 71% white	114 infants (53 girls) M=14.71 months, SD=.46 Only 14 month data collection used	14-Months *Prohibition task *3-boxes *Ball Run	*Each test in battery assumed to be unidimensional, model with each task loading onto separate factor with CFA as a good fit *none of the EF tasks were related r<.04
Devine et al., (2019)2	East of England, United Kingdom 92.8% mothers and 94.9% fathers White British 84.6% mothers 77% fathers had undergraduate degree 60.8% mothers and 61.4% fathers from professional occupations	196 infants (87 girls) M=14.42 months, SD=.59 Longitudinal EF at 1 timepoint	14-Months *Prohibition task *3-boxes *Delayed response *Ball Run	*Each test in battery assumed to be unidimensional, with each task loading onto separate factor with CFA as a good fit *Prohibition was weakly associated with 3-boxes (.27*) and ball run (.25*) when examining latent factor covariances. There were no other significant associations.
Frick et al., (2019)7	Uppsala, Sweden Full sample (n=66) 86% mothers and 80% fathers had university degree	66 infants (35 girls) M=12.47 months, SD=.43 M=18.51 months, SD=.73 M=24.60 months, SD=.61 Longitudinal at 12, 18, 24, 36, and 74 months ns ranged from 58 - 66 on tasks	12-, 18-, and 24-Months *Prohibition tas *Hide and Seek Task	*Prohibition was not significantly related to hide and seek at 12, 18, or 24 months, spearman’s rho <.00 *Prohibition at 12, 18, and 24 months was correlated *WM was not correlated at 12, 18, and 24 months *no correlations within the hide and seek were found
Hendry et al., (2022)	Oxford, United Kingdom 85.2% White 93% undergraduate degree or higher	108 Infants at time 1, 73 infants at time 2 M=10.18 months, SD=.38 M=16.07 months, SD=.42 Longitudinal at 2 timepoints	10 & 16 Months *Toy Prohibition *Touchscreen Inhibition *A-not-B *Early Childhood Inhibitory Touchscreen Task (ECITT)	10-Months *Touchscreen prohibition was related to touchscreen approach, r=.31*, and toy prohibition, r=.35* *No other correlations were significant 16-Months *Toy prohibition was related to touchscreen prohibition, r=.59 *A-not-B switching was related to ECITT Accuracy difference, r=.45*, and ECITT switching, r=.31* *Children generally improved from 10 to 16 months, with the exception of the accuracy on the ECITT.
Holmboe et al. (2018)	104 infants (53 girls) Greater London Area, United Kingdom 74% White Predominantly higher SES 62.4% at least 16 years education	9 months (within 1 week) Longitudinal, EF with multiple measures assessed at 9 months	9-months *Freeze-Frame Task *A-not-B Task	*Individual task performance considered *A-not-B at 9 months related to Freeze Frame Performance, r=−.40 *Relations were replicated in a SEM
Hughes et al., (2020)2	East of England, United Kingdom New York State, United States Netherlands 84.3% at least undergraduate degree	422 infants at time 1 404 infants at time 2 M=14.42 months, SD=.57 M=24.47 months, SD=.78 Longitudinal EF at 2 timepoints	14-Months *Prohibition task *Multilocation Search Task *Ball Run 24-Month *Baby Stroop *Multilocation Search Task *Ball run	14-Months *reliability coefficient alpha=.37 (based on tetrachoric correlations) 24-Months *reliability coefficient alpha=.58 (based on tetrachoric correlations) A single aggregate score based on adding the number of EF tasks passed was considered.
Johansson et al., (2016)7	Sweden 92.5% at least 1 parent had university degree	66 infants (35 girls) M(=371days, SD=13days Longitudinal at 12 months, 24 months (2 tasks), and 36 months	12-Months *Hide-and-seek *Reverse categorization *Prohibition task 24-Months *Spin the pots *Reverse categorization	12-months *no significant correlations between EF tasks *Individual scores for each task considered 24-months *no significant correlations between EF tasks 12 month EF did not correlate with 24 month EF
Li, Devine et al., (2022)8	Netherlands China Incomes 16% higher than average national level Majority of parents had bachelors degree or higher	124 Dutch infants (55% girls) M=14.19 months, SD=.52 63 Chinese infants (49% girls) M=14.83 moths, SD=1.15 Longitudinal EF at one timepoint	14 Months *Prohibition Task *Multilocation Search Task *Ball Run	*Latent factor scores created in SEM for working memory, cognitive flexibility, and inhibition *There were no significant correlations among 3 EF scores
Li, Emmen et al., (2022)8	China 75% mothers and 75.5% fathers had bachelor degree or higher Most families had middle-to-upper income levels, 16% higher than average level in Shenzhen, China	63 infants (49% girls) M=14.83 months, SD=1.16 Longitudinal EF at one time point	14 Months *Prohibition Task *Multilocation Search Task *Ball Run	*Latent factor scores created in SEM for working memory, cognitive flexibility, and inhibition * 3 components scores of EF were unrelated
Lui et al., (2021)	United Kingdom 83.44% White M=17.9, SD=3.15 years maternal education, M=17.3, SD=3.26 years maternal education,	163 infants (78 girls) M=305 days (approximately 10.2 months), SD=6.63 Longitudinal study only 10 month assessment considered	10 Months *Toy Prohibition *Early Childhood Inhibitory Touchscreen Task (ECITT)	* ECITT performance did not significantly correlate with toy prohibition, r=.15 *Individual task performance considered
McHarg et al., (2020)2	New York City, United States (N=100) England, United Kingdom (N=194) Netherlands (N=122) 81.5% mothers and 72.7% fathers had bachelors degree or higher	416 infants (204 girls) M=14.42 months, SD=.57 Longitudinal EF at one time point	14-Months *Prohibition task *3-boxes *Ball Run	*Each test in battery assumed to be unidimensional, each task loaded onto separate factor. *Correlations between factors were not reported.
Miller & Marcovitch (2015)	Southeastern United States 38% Caucasian, 32% did not respond 32% reported average income above $60,000, 34% did not respond	47 infants (22 girls) M=14.38 months, SD=.34 M=18.48 months, SD=.35 Longitudinal EF at 14 and 18 months	14-Months & 18-Months *A-not-B with multiple locations *Forbidden Toy *Three Boxes *Imitation Sorting	14 Months *Imitation sorting related to forbidden toy, r=.32* *No other correlations were significant 18-Months *Performance across tasks was not correlated *Only IST and number of EF tasks passed showed longitudinal stability from 14 to 18 months *Task performance increased from 14 to 18 months on all but forbidden toy task *individual scores for each task considered along with number of EF tasks passed
Nichols et al., (2005)	Major metropolitan area, United States Parents middle- to upper-class and had at minimum graduated from high school 91% Caucasian	39 infants (16 girls) n=39 at 14 months n=35 at 16 months n=33 at 18 months	14-, 16-, & 18-months *Means End *Delay nonmatching to sample	*slope (growth) was considered for means end and delay nonmatching to sample tasks *relations between tasks not considered
Ribner et al., (2022)2	Original Sample (n=484) New York City, United States (N=131) England, United Kingdom (N=221) Netherlands (N=132) 89.2% mothers and 92.5% fathers White 84.3% attained undergrad degree or higher	Eligible Sample 404 infants (195 girls) M=14.42 months, SD=.57 M=24.47 months, SD=.78 Longitudinal EF at 14 and 24 months	14-& 24-Months *Prohibition task *Multi-location search task *Ball Run Task	14 Months *Each task loaded onto separate factors. *Components were separable indicated by weak and non significant correlations. 24-Months *Exploratory factor analysis indicated 3-factor model was best fit, indicators from each task loaded on latent variable representing the task. *Working memory related to inhibition and shifting, rs=.21** *Factors representing shifting were correlated across time, r=.14.
Sun et al., (2009)9	Brisbane, Australia 37 Very Preterm Infants (48.6% female) 74 Full Term Infants (48.6% female)	Preterm M=28 weeks gestational age (7 months) SD=1.9 Full-term M=39 weeks (9.75 months) SD=1.1 EF assessed at 1 point for 2 samples	7- (gestational) & 9.75-months (full term) *Modified A-not-B *Planning Task	*Individual scores for each task considered *Differences between preterm and full term were found on EF tasks, but relations between tasks not considered.
Sun & Buys (2012)9	Brisbane, Australia 37 Very Preterm Infants (48.6% female) 74 Full Term Infants (48.6% female)	Preterm M=28 weeks gestational age (7 months) SD=1.9 Full-term M=39 weeks (9.75 months) SD=1.1 EF assessed at 1 point for 2 samples	7- (gestational) & 9.75-months (full term) *Modified A-not-B *Planning Task	*Individual scores for each task considered *Most EF measures did not correlate with each other.
Wiebe et al., (2010)	30 infants (16 girls) Midwestern city, United States Primarily Caucasian	Time 1=15 months, (14;24-15;11) Time 2=20 months 14 days (20;08-20;23) Longitudinal at 2 time points	15- & 20-Months *A-not-B with invisible displacement *3-boxes tasks *Immediate imitation *Imitation with distractors *Planning	*individual scores for each task considered (2 per tasks that did not show floor effects) 15-Months *Planning related to A-not-B, r=.39* and immediate imitation, r=.36* 20-Months *A-not-B related to imitation with distractors, r=.38, planning related to immediate imitation, r=.33* *Only imitation with distractors, r=.36*, and immediate imitation r=.44* showed longitudinal stability *Task performance generally increased from 15 to 20 months on all tasks

Note:

^*p<.05,

^**p<.01,

EF = executive function PCA = principal component analysis, DoG=Delay of Gratification, CFA = Confirmatory Factor Analysis, DCCS = Dimensional Change Card Sort, IST=imitation sorting task SEM = structural equation modeling, PCA = principal component analysis

¹indicated overlap in sample

²indicated data from the same large longitudinal international sample of families in 3 countries

³both noted participants were recruited as part of an ongoing longitudinal investigation of cognition and emotion development but did not specify they had overlapping data

⁴authors noted that samples overlapped (verbal correspondence)

⁵noted study was derived from an ongoing longitudinal study, but did not note if there was overlap in samples

⁶noted children participated in part of an ongoing longitudinal study, but did not note if there was overlap in samples

⁷noted overlapping sample

⁸noted participants were part of a cross-cultural longitudinal study but did not note if there was overlap in samples

⁹noted overlapping sample

2.0. Universality in the Emergence of EF

It is common in the study of EF and its development to view EF as a domain-general process related to cognitive control that is important to development across domains (e.g., social cognition, school readiness and competence, social relationships, for examples see Blair & Razza, 2007; Carlson et al., 2004; Miller et al., 2020). As such, some of the questions that have surrounded the development of EF have been related to better understanding how to conceptualize this domain-general ability by focusing on questions related to the structure, growth, and longitudinal stability across the lifespan. This has also come with an emphasis on methodology, as the integration of multiple measures across contexts and ages are likely necessary for the study of a domain-general ability that operates in a variety of settings (i.e., related to the task impurity problem). In the section that follows, we consider what the studies from the present scoping review may suggest in terms of understanding this potentially domain-general ability in cognitive development.

2.1. Ages Considered

Before reviewing patterns across studies, it is important to understand what currently exists in terms of the ages tested and tasks used to examine EF within the first few years of life. Overall, the average age ranged from approximately 9 to 26 months. Twenty-four months was a common starting point of assessment (+/− 3 months, 17 published studies). Assessment of EF via a battery of tasks at this age is not particularly novel, as several researchers have used an EF battery dating back at least a decade (Bernier, Carlson, Bordeleau, et al., 2010; Carlson et al., 2004). This is in line with many studies conducted during the preschool years (e.g., Wiebe et al., 2008, 2011). There were fewer studies with samples starting at approximately 18 months (+/− 3 months, 7 published studies), though studies starting at approximately 12 months were common (+/− 3 months, 17 published studies). Further, the majority of the studies published with younger samples have been conducted in the past 5 years, with the exception of a few studies published earlier (Diamond et al., 1997; Johansson et al., 2016; Miller & Marcovitch, 2015; Nichols et al., 2005; Sun et al., 2009; Sun & Buys, 2012; Wiebe et al., 2010). Further, the earliest assessment using a battery of tasks to measure EF was approximately 9 months of age (Holmboe et al., 2018). This may suggest that the use of a battery to examine EF is a relatively new approach for children in the first 2 years of life (specifically before 24 months), though it seems to be growing in popularity within the past 5 years.

2.2. Tasks Utilized in an Early EF Battery

Table 2 summarizes common tasks used within early EF batteries classified by the EF component it is commonly cited as assessing. Though it is important to note that not all tasks are classified in the same way by all researchers and it may be difficult to isolate specific EF components given that inhibition may be necessary for most, if not all EF tasks (e.g., any task that involves switching/shifting a response will also involve inhibition of a response, see Miyake & Friedman, 2012). For tasks used to assess inhibition, children were commonly asked to delay gratification and wait for an attractive item (e.g., toy, snack) with assessments ranging from the earliest assessment at 9 months to latest assessment at 26 months. These types of tasks have been classified as simple response inhibition requiring the withholding or delay of a prepotent or automatic response (Garon et al., 2008). Delaying gratification tasks have also been discussed as involving more “hot EF” which refers to higher order cognition that operate in situations that involve motivation or emotion (Zelazo & Carlson, 2012). There were several other inhibition tasks (i.e., the freeze frame, walk-a-line slowly, tower, Early Childhood Inhibitory Touchscreen Task) that are also likely considered simple inhibition tasks that involved inhibiting a prepotent motor response (e.g., looking, walking fast, pressing on a screen). There were only two tasks (i.e., the tricky box and Simon says) that may assess more complex inhibition, that involves holding and responding to a rule in mind in addition to inhibiting a prepotent response (Garon et al., 2008).The tricky box or detour reaching involves not only inhibiting a prepotent response of reaching for a toy visible through plexiglass, but also holding information in mind that will assist children in acquiring the toy (e.g., remembering that the plexiglass window had to be opened by pulling a knob, Marciszko et al., 2020). Although this task was used to assess complex inhibition in 18-month-olds, there are few other potential complex inhibition tasks that have been used within an EF battery at this age. The one other task—the Simon Says task—has only been administered with the oldest age and was removed from the model because it loaded negatively to the EF factor score (Broomell & Bell, 2022). Thus, more studies examining inhibition in tasks outside of a delay of gratification paradigm may be useful to better understanding how to best assess EF at this age via an EF battery, as some have suggested that the demands of the task like the Simon Says task could be too difficult for children this young (Broomell & Bell, 2022; but see Garon et al., 2008; Kochanska et al., 1996).

Table 2.

Summary of tasks used in EF batteries in children 26 months of age or younger

Task 1) Potential Inhibition	Ages	Description
Delay Task
Delay of Gratification Snack Delay, Tongue task	26M, 25M, 24M, 22M, 21M, 18M	Children were asked to wait to eat a snack either placed in front of them, under a cup, or on their tongue.
Crayon Delay, Externally Imposed delay, Gift-in-Bag, Prohibition task, Simple inhibition, Touchscreen inhibition, Wrapped Gift	24M, 22M, 18M, 16M, 14M, 12M, 10M	Children were asked to wait to touch an attractive item (e.g., box of crayons, toy, attractive item on touchscreen) or wait and not peek at a gift that was being wrapped.
Simple Inhibition
Freeze Frame	9M	Children watched animated cartoon on the center of the screen and the animation was frozen if they looked to a distractor square on the right or left of the screen.
Walk-a-Line-Slowly	22M	Children were asked to walk down a line to and from mother as slowly as possible.
Tower	22M	Children were asked to take turns with experimenter building a tower together.
Early Childhood Inhibitory Touchscreen Task	16M, 10M	Children were asked to press a smiley face (rather than a blank button) on a screen and were reward with an animation. Trial presentation was randomized with constraints, with the smiley face appearing on one side 75% of the trials (prepotent trials) and the other side 25% of the trials (inhibitory trials).
Complex Inhibition
Simon Says	24M	Children were asked to follow the instructions of one “nice puppet” but not to do what a “mean puppet” says.
Tricky Box, Detour reaching	18M, 17M	Children were asked to get a toy placed inside a box with plexiglass window that had to be opened by pulling a knob.
2) Potential Working Memory
Search Tasks
3-boxes	20M, 18M, 15M, 14M	Children were asked to find toys hidden simultaneously in 3 different boxes one at a time and update their search based on prior search.
Spin the pots	26M, 25M, 24M	Children were asked to find 6 stickers in 8 different pots that were rotated.
Hide and Seek, Find the toy	18M, 12M	Children were asked to search until they found a toy hidden in one of 6 drawers. Toys were hidden in a new location each trial (4 trials total).
Imitation Tasks
Imitation Sorting	14M, 18M	Children were asked to imitate sorting an increasing number of toys into 2 buckets.
Immediate Imitation	20M, 15M	Children were asked to imitate a multistep sequence to complete an action.
Imitation with distractors	20M, 15M	Children were shown two multistep sequences. They were presented with all materials for one action and partial tools for another action, and they were asked to imitate the multistep sequence for which they had the materials for.
3) Potential Flexibility or Shifting
Search Tasks
Multilocation Search Task, A-not-B with multiple locations	24M, 21M, 20M, 18M, 16M, 15M, 14M, 10M	Children were asked to find a toy in one location multiple times correctly (location A), then they are asked to find the toy in a new location (location B).
Spatial Reversal	21M	Children were asked to find a toy that was hidden out of view to children at one location (A). After they found it 4 times at location A it was moved to a new location (B)
Delayed Nonmatching to Sample	14M, 16M, 18M	Children were asked to participate in training in which sample covered a reward. After delay, sample was presented again with new sample and reward was under the new sample. Children had to displace new sample to receive reward.
Delayed response	17M, 14M	Children were asked to find toy hidden in one of 2 locations after delay with object hidden in a fixed order alternating between left and right.
Rule Following Tasks
Shape Stroop	26M, 25M, 24M, 22M	Children were asked to point to a small fruit embedded in a larger fruit.
Baby Stroop	26M, 25M, 24M	Children were asked to reverse the rule, where they fed a mommy with a smaller spoon and baby with a large spoon (after they learned to feed the mommy doll with the large spoon and baby with the small spoon).
DCCS, Shape and Color Sort (Modified DCCS)	24M	Children were asked to sort cards (red or blue car or flower) according to shape or color. They were then asked to sort by the opposite. Pre-switch measured. Modified DCCS: Children were asked to sort items by shape then color (ignoring shape).
Reverse Categorization	25M, 24M, 18M, 12M	Children were asked to put large blocks in a big box, and small blocks in a small box. Then the rules were reversed.
Ball run task	24M, 14M	Children were asked to learn a rule (e.g., place a green ball in green hole when green, yellow, and red holes were visible) to activate a toy. If children scored 4+ of 6 trials correctly, children moved to the reversal phase where they placed a different color ball in a new hole (e.g., red ball in the red hole) to activate the toy.
4) Other (Planning or Problem Solving)
Means End	14M, 16M, 18M	Children were asked to complete trials assessing the use of objects and relations between objects as means (e.g., use stick as means).
Planning	10M	Children were asked to overcome 1-3 obstacles to obtain an object 1) pull cloth to reach object 2) remove barrier before pulling cloth and 3) remove barrier, pull cloth to bring string into reach, and pull string to reach object

For tasks used to assess working memory, different types of search and imitation tasks have been used to examine children’s ability to hold and update or manipulate information in mind. Most of the search tasks involve hiding toys in multiple locations to examine how well children hold in mind where toys are hidden and update their searching. These tasks have been used across a broad age range within EF task batteries from 12 to 26 months and align with Garon and colleagues (2008) review of tasks used to assess complex WM at this age (i.e., involving both holding and updating information in mind). Imitation tasks involve asking children to hold an increasing number of items in mind (e.g., sort items into buckets or remember actions to imitate a multi-step sequence). These tasks may rely more heavily on simple WM, which involves primarily holding information in mind over a delay and not necessarily manipulating and updating this information. However, some variations of the tasks, such as the imitation with distractors may involve holding information in mind in addition to inhibiting irrelevant information. These tasks have been used to assess WM in children from 14 to 20 months of age.

The flexibility or shifting tasks either used a search or rule-following paradigm to examine children’s flexible behavior in their ability to shift to a new response. For the search tasks, children usually built a response to search for an object at one location and then behavior was examined when the object was moved to a new location and children had to shift searching behavior to a new location. These types of tasks have been examined in an EF battery in children from 10 to 24 months and are thought to assess response switching, which is specifically focused on children’s ability to switch to a new stimulus-response set (searching in a new location, Garon et al., 2008). The other tasks involve rule following, but the focus is typically on children’s ability to shift attention within the task rather than a physical response. For example, in the dimensional change card sort task (DCCS, Zelazo, 2006), children sort stimuli by one dimension (e.g., a red square by color to match a red triangle rather than blue square). Then children are asked to shift their attention from sorting by color to sorting by shape (e.g., sorting the red square to the blue square rather than red triangle). Thus, the shift comes in children’s attention to the new aspect of the stimuli. Although these tasks have been administered with children ranging from 12 to 26 months, most are administered with children closer to 2 years of age. When task were administered earlier, most children younger than 2 years of age performed very poorly on the switching task (considered to be the most complex because it integrates multiple subcomponents of EF, Garon et al., 2008) likely indicating the task was too difficult (e.g., Johansson et al., 2016, average of 1.88 out of a possible 12, range 0-4).

Finally, there were several tasks that were not easily classifiable (i.e., the means end and planning task) that seemed to draw on general problem solving or planning (Nichols et al., 2005; Sun et al., 2009; Sun & Buys, 2012). Although it is likely that performance on these tasks draw on EF (e.g., Zelazo et al., 1997), they do not appear to target any of the potential underlying EF components (i.e., WM, inhibition, or shifting). Taken together, the selection of tasks included in task batteries for toddlers has a few distinctions that are worth discussing. First, unlike tasks with older children and adults many of the tasks rely on training behavior and physical responses, likely due to children’s limited linguistic ability at this age. The tasks that do rely heavily on language are typically administered successfully only with the oldest age group (24 +/− 3 months). Second, the types of tasks that are administered with the younger age groups could be limited in the complexity of EF they can assess. The tasks reviewed in this paper broadly align with Garon and colleagues (2008) integrative EF model suggesting that EF builds in complexity as new skills are integrated in children’s cognitive systems. Existing batteries of EF for toddlers seem to primarily draw on assessing the simpler (and less integrative) measures of inhibition, WM, and shifting. Although there are a few measures of what Garon and colleagues term complex inhibition, WM, and attention-shifting, most are administered to children closer to 24-months of age (with the exception of WM) with limited abilities in children this young. Not everyone classifies EF tasks in the same way, so it may be useful for EF batteries to consider what element of EF each task may be tapping into and how this may impact our understanding of EF development within the first 2 years.

2.3. Structure of EF based on EF task batteries

One advantage to using a battery of EF tasks is that researchers can examine the structure to better understand what the assessment across multiple tasks can tell us about EF (i.e., whether we should assess separate components or simple vs. complex forms of EF, does a unitary EF construct better represent the data?). A common way to address this issue is to examine the relationships between EF task performance. Here, the data shows several distinct patterns. For the earliest assessments 12 months (+/− 3 months) most studies showed very few correlations and low reliability among tasks (see Table 1). Tasks that had sample sizes large enough to examine factor structure found that the best structure that fit the data was to have tasks load onto their own separate factor with each test in the battery assumed to be unidimensional. Although some authors labeled the factors by their component of EF they were thought to assess (i.e., WM, flexibility, and inhibition), the complexity of the components were not further tested and there were not multiple tasks that allowed for extracting latent factors for each EF component based on multiple task performance. Although some authors considered an aggregate score of EF (e.g., based on the number of tasks passed), it was common for authors to assess individual task scores or a combination of the two (i.e., consider task scores separately but also creating an overall EF score based on number of tasks passed).

The age that has the fewest assessments was 18 months (+/− 3 month), however the data for this age looks similar to the 12-month assessments. There were few correlations between tasks for children this young. In addition, it was common to consider individual task scores in addition to number of EF tasks passed.

The structure at 24 months (+/− 3 months) may suggest a bit more cohesion between tasks, as the reliability coefficient between tasks increased by approximately .20 from 14 to 24 months (Hughes et al., 2020) and there were more significant correlations between tasks at this age. Studies that examined the factor structure suggested that performance across some tasks may reduce to factor components. For example, studies by Bernier and colleagues suggested that a 2-factor structure may reduce to impulse control and working memory (see Table 1, Bernier, Carlson, Bordeleau, et al., 2010; Bernier, Carlson, & Whipple, 2010; Bernier et al., 2012, 2014), suggesting that at this age there is some overlap between performance on multiple WM tasks. However, other studies at this age suggest a single EF factor (Broomell & Bell, 2022) or EF composite (Carlson et al., 2004; Joyce et al., 2016; McHarg et al., 2020) was most appropriate, while others found little evidence for relations among tasks when considered separately (Cheng et al., 2018; Gago Galvagno et al., 2019, 2021; Pauen & Bechtel-Kuehne, 2016; Zhang & Wang, 2022). Thus, work suggests stronger relationships between task performance and a potential emerging unitary structure may be apparent at this older age, though not for all studies. This could be due in part to differing measures used across EF batteries and differences in the ages examined, as children ranged from an average of 21 to 27 months in the oldest age group, which could be a meaningful difference in EF that is likely rapidly developing across this time.

However, there are limitations to studies within the first two years of life that restrict the ability to address early EF structure. The focus on the underlying structure of EF emerged from adult EF work. This approach involves administering several tasks for each latent factor so that researchers could use confirmatory factor analysis or structural equation modeling to better understand how EF is organized, address the task impurity problem, and understand how EF plays a role in complex cognition (Miyake et al., 2000). This has inspired developmentalists to ask similar questions about the developing structure of EF at younger ages (e.g., Lehto et al., 2003; Wiebe et al., 2008, 2011). However, as can be seen from Table 1, most studies in early EF do not approach the structure of EF with the same focus on latent factors. This is likely due to the small sample sizes and limitations in available instruments for younger preverbal children. One successful approach to data collection challenges has been the formation of international research teams that collect data in multiple research sites across the globe (e.g., Blum & Ribner, 2022; Devine et al., 2019; Hughes et al., 2020; McHarg et al., 2020; Ribner et al., 2022), though it is important to note that multiple studies that use a factor analytic approach in Table 1 acknowledge that they use overlapping data from the same international sample. Thus, more work is needed to better understand developing EF structure within the first couple of years of life.

2.4. Growth and Longitudinal stability based on EF task batteries

Many studies have employed a longitudinal design to examine EF development measured via a task battery, although they vary in the ages of assessment. Eight studies had longitudinal data with measurements in the first 2 years of life assessed via an EF task battery. Of the studies that examined growth, all indicated that EF generally increased from early assessments to later in the second year (Hendry et al., 2022, from 10 months to 16 months; Miller & Marcovitch, 2015, from 14 months to 18 months ; Wiebe et al., 2010, from 15 months to 20 months, see Table 1). Evidence for longitudinal stability suggested that stable individual differences were less evident. Several studies have found no relationship between EF assessed earlier in the second year and later assessments at 18 (Hendry et al., 2022) and 24 months (Johansson et al., 2016). Some studies have found partial evidence for longitudinal stability, with typically only one or two measures from the battery demonstrating a relationship between earlier and later performance across the second year (Frick et al., 2019; Miller & Marcovitch, 2015; Ribner et al., 2022; Wiebe et al., 2008). However, when researchers assessed all tasks together by considering the number of tasks passed there was evidence that the number of tasks passed earlier in the second year of life was related to number of tasks passed at 18 (Miller & Marcovitch, 2015) or 24 months (Hughes et al., 2020).

It is important to note that other studies included data from longitudinal data sets capable of assessing growth and longitudinal stability beyond 24 months. This work has shown that EF assessed at 24 months does predict EF at later assessment ages. For instance, several studies have demonstrated that some EF assessed via a battery of tasks at 24 months related to EF assessed later in the fourth year of life (e.g., Bernier et al., 2012; Broomell & Bell, 2022; Carlson et al., 2004; Cheng et al., 2018; Joyce et al., 2016) and single EF assessments in the first year have been shown to relate to EF assessed at year 3 (Blankenship et al., 2019). The fact that assessments at 24 months appear to be more predictive of later EF algin with other early EF findings demonstrating dramatic improvements and more cohesion between performance on EF tasks at this older age (e.g., Carlson et al., 2004; Hughes et al., 2020). An important question for future research will be to better understand why we do not see the same longitudinal stability in the first couple years of life that is typical of assessments later in development. There are fewer data sets to draw on that can address this question before 24 months of age that use a battery of EF tasks. It is possible that the lack of longitudinal stability may be due to methodological issues (e.g., ceiling or floor effects, dramatically different measurements across age ranges), though several of the studies use similar or identical tasks and demonstrate variability in performance across multiple age ranges. More work is needed to determine if the lack of longitudinal stability is yet another unique EF finding evident within the first two years that could be indicative of a distinct period of EF development.

3.0. Context Specificity in the Emergence of EF

Although section 1 suggests some potential universal developments related to emerging cohesion, significant growth, and potential longitudinal stability across the first years of life in EF, this does not exclude the possibility of context-specific developments grounded within children’s sociocultural context. Further, some of the inconsistent findings we see across studies could also be related to variability in children’s contexts in which EF is measured. In this second section, we review several avenues for examining development across different elements of children’s sociocultural contexts to better understand the potential for context-specific development.

3.1. Examinations across country and racial/ethnic backgrounds

Even though the study of EF using a battery of tasks seems to be a somewhat novel approach to understanding early EF development, there is diversity across the participants that have been part of the work to date. Within North America, testing was conducted across different regions of the United States (US) and Canada, though it is important to note that many of these studies were conducted in metropolitan areas and the samples were predominantly White with at least a middle-class socioeconomic status (when noted). In Europe, testing was conducted in Germany, Sweden and the United Kingdom (UK). When noted, the samples from these countries seemed to come from at least middle class and were primarily White. Within Asia, testing occurred in China and worked with samples from mid to upper income levels (when noted). Within South America, testing occurred within Argentina and worked with samples that were of low to mid socioeconomic status (SES) and identified as Argentine. Finally, several empirical studies examined early EF from multiple countries within one study with samples including participants from the US, UK, and Netherlands (Braren et al., 2021; Hughes et al., 2020; McHarg et al., 2020; Ribner et al., 2022) and the Netherlands and China (Li et al., 2022). For the one study examining children from the Netherlands and China, Dutch infants showed higher inhibition than Chinese infants, which conflicts with later differences observed between children from Chinese and Western cultures (Sabbagh et al., 2006; Wang et al., 2016). The authors suggest this may be because children are not yet in school, where the early maturity typically seen in Chinese school-age children may be due to differences in educational structure and teaching (Li et al., 2022). For studies including samples from the US, UK, and Netherlands, attempts were made to account for systematic differences as a result of country (e.g., Ribner et al., 2022) but understanding different pathways in development that may emerge between countries were generally not explored. It was noted that this would be a useful consideration for future work (McHarg et al., 2020), but the samples recruited within these studies were still largely homogenous consisting of Western, industrialized, rich, and highly educated participants (e.g., Ribner et al., 2022).

Importantly, despite the diversity of the samples, the patterns of early EF development discussed in section 1 seemed to hold. For example, authors of studies from non-Western samples have noted how the lack of coherence and stability within measures of early EF mirror what is seen in other studies, suggesting the first 2 years as a distinct period of development and period of significant growth (Cheng et al., 2018). This is also discussed by Gago Galvagno and colleagues (2021) as their work suggests performance across tasks measured with a battery of tasks in the first 2 years of life do not correlate well but the strength of these relations and EF performance improves into the third year within their examination of a Latin American sample with low to mid-SES. The authors note that this may suggest potential universalities with the first few years being a period of emergence in EF development, while also acknowledging that other sociocultural factors may be important to understanding how contextual differences may play a role in EF’s foundation and emergence (e.g., how bilingualism may reflect unique strengths built within sociocultural environments).

However, it is also important to note that more work is likely needed in this area as there are several limitations to consider. For one, many of the studies discussed come from the same labs and although there are several published studies with diversity in the samples examined, some of the studies rely on either the same longitudinal study or are conducted in the same regions (see Table 1). In addition, an important limiting factor was the requirement that studies in the present paper were published in English (though there were some studies published in multiple languages, e.g., Gago-Galvagno et al., 2022). Thus, as noted by several of the authors, although several patterns seem to emerge across samples related to a lack of (but improving) coherence between task performance, dramatic improvements across the first couple years of life, and tenuous longitudinal stability indicative of a unique and emerging period of EF development, more work is needed especially with diverse samples to understand potential universals in early EF development and points for context specific development.

3.2. Examinations across SES

Another important consideration in the diversity of our sample is SES. As noted in the prior section, despite the diversity in testing sites most samples were from middle to upper SES backgrounds in terms of income and education. The examination of EF in the context of poverty has been identified as an important research direction, as associated stress potentially relates to less reflection and flexibility in responses (Blair, 2016; Blair et al., 2011). However, most studies have examined SES with older samples. Of the studies that examined SES within the first 2 years, there was a similar pattern suggesting that higher SES was associated with better EF for some measures (Bernier, Carlson, Bordeleau, et al., 2010; Broomell & Bell, 2022; Gago Galvagno et al., 2019, 2021; Gago-Galvagno et al., 2022). However, not all studies found effects of SES. Gago Galvagno and colleagues (2022) found that when multiple factors related to the home environment were added into a regression analysis to predict the number of EF tasks passed (i.e., type of housing, overcrowding, frequency of reading, number of books, cell phone use, computer use, internet use, gender, and age) the model accounted for a significant portion of the variability (37%), but the predictors were not the elements more typically related to SES (e.g., type of housing, overcrowding). Once multiple factors were in the model, it appeared that children who read more and used cell phones less tended to have better EF.

These results are in line with recent commentaries (Raver & Blair, 2020) on EF suggesting that it is necessary for EF research to place the examination of EF development in context considering not only EF associations with poverty, but to center research more fully in the broader sociocultural environment. This may include considering the structural social and economic inequalities communities may face in our models to better understand how the stressors related to poverty are part of the broader social systems children and communities exist within. Further, it is important to consider the different contexts in which developing children are asked to exercise regulation and draw on families and communities to understand the goals, expectations, knowledge, and skills they bring (Moll, 1992). Thus, more work is needed to better understand EF in the context of poverty at this age from a broader sociocultural context that also draws on understanding the strengths and resilience of communities that face structural inequities (Raver & Blair, 2020).

3.3. The role of communication and language

For EF research within the first two years, two factors within the sociocultural environment seem to emerge prominently. The first consistent factor within early EF studies— language and communication—is not a new consideration in EF developmental work. The proposal that language and communication developed within one’s social and cultural environment plays a role in the way individuals think about the world and control of behavior dates back to the classic works of Vygotsky and Luria (Luria, 1979; Vygotsky, 1986) and has a wealth of research investigating a link between the two abilities at older age ranges (for a recent review, see Bruce & Bell, 2022). The data from the few studies that measure EF and verbal abilities in the first few years of life suggest the relationship may be tenuous. For example, although Bernier et al., (2010) found concurrent correlations between expressive verbal ability and performance on individual EF tasks at 18 and 26 months, several studies only found language-EF links for one EF task (Poulin-Dubois et al., 2022; Zhang & Wang, 2022) and Miller and Marcovitch (2015) did not find any evidence for a significant concurrent language-EF relationship at 14 or 18 months when examining individual task performance. Miller and Marcovitch did however demonstrate a longitudinal relationship, where parent-reported verbal comprehension at 14 months positively predicted the number of EF tasks passed at 18 months. Similarly, Carlson et al. (2004) demonstrated a language-EF relationship when EF was considered as a unitary score at 24 months. Finally, several studies failed to find a concurrent language-EF relationship when examined at approximately 14 months (Ribner et al., 2022) and 2 years of age (Joyce et al., 2016; McHarg et al., 2020) and Poulin-Dubois et al. (2022) did not find evidence for a benefit of bilingualism to EF that is commonly found in older ages.

One reason for the inconsistent results at this age may be that linguistic abilities in the first 2 years of life are limited and rapidly developing. Further, tasks at this age deliberately attempt to minimize linguistic demands, thus potentially also minimizing the relationship between language and EF. Another skill or construct that emerges early in the first few years of life, has strong ties to later language development, and has been proposed as potentially a more age appropriate measure of communication and representation for younger samples is joint attention (Miller & Marcovitch, 2015). There are different types of joint attention (i.e., shared reference to an object or event), with the most assessed being responding to joint attention (i.e., RJA related to following others attention) and initiating joint attention (i.e., IJA related to directing the attention of another). The studies that have examined joint attention have found evidence for EF linking to RJA and some links to IJA. For example, Gago Galvagno et al., (2019) showed that higher scores in RJA and IJA related to more EF tasks passed for infants between 18 and 24 months. However, in a later study Gago Galvagno et al., (2021) showed that more instance of RJA (but not IJA) related to better EF performance on individual tasks and RJA at 21 months. Miller and Marcovitch (2015) found a similar pattern that RJA (but not IJA) positively related to one EF task and the total number of EF tasks passed at 14 and 18 months (though the 18-month relationship was marginally significant). Marciszko et al. (2020) also found that more gaze following (related to RJA) at 6 months was related to better simple inhibition but not complex inhibition or WM at 18 months (IJA was not assessed). There have been several explanations for the early RJA-EF relationship, with some suggesting that RJA provides a measure of early social cognition (i.e., sharing attention through gaze is related to creating internal models of others’ actions) which helps infants track and understand goal-oriented behavior in others that aids in their own inhibitory control development (Marciszko et al., 2020). Others have suggested that RJA may not require sophisticated social cognition (at least when measured with older infants), rather it may refer to the ability to orient attention that may help explain why at a time of rapidly developing and less cohesive EF abilities, children may rely on more basic selective attention abilities to guide behavior likely reflected in RJA. In other words, children with better RJA may perform better on some EF tasks because they are more attentive and responsive in the testing session (Miller & Marcovitch, 2015).

This leads to the other element of joint attention—IJA. At first review, it does not seem like there is much evidence for an IJA-EF relationship when measured concurrently. However, despite failing to find concurrent relationship between IJA and EF, Miller and Marcovitch (2015) were able to find a longitudinal relationship with Higher IJA behaviors (i.e., active directing of attention involving pointing and showing) at 14 months predicting EF tasks passed at 18 months. The authors suggested that this finding aligns with the assertion that abilities related to representation (e.g., pointing requires children to essentially label in item to share attention to, Zelazo, 2004) are important to maintaining task-relevant information—a key ability important to EF (e.g., Miyake & Friedman, 2012) that is likely developing within one’s sociocultural environment toward the end of the second year (Wiebe et al., 2010) and may be responsible for the steady improvement in early EF. This also aligns with representational models of EF development suggesting that the transition to more sophisticated EF relies on the children’s abilities to form, represent, and maintain task-relevant information (Marcovitch & Zelazo, 2006, 2009; Zelazo, 2004).

The current literature suggests the relationship between language and EF is complex, as relationships are not consistent with variations across age, tasks, and measurement approaches to EF (e.g., longitudinal vs. concurrent, separate tasks vs. composite or factor EF score). Further, language and communication are multifaceted. From a sociocultural perspective, communication and language is an important element of the environment that varies by culture (e.g., Akhtar & Gernsbacher, 2007, 2008; Hoff, 2006) and influences the way we think (Luria, 1979; Vygotsky, 1986). However, linguistic ability (e.g., vocabulary) is also often studied as a cognitive domain or ability that may partially relate to intelligence, given the strong correlations and inclusion of linguistic ability in intelligence testing (e.g., Hodapp & Gerken, 1999). Current work examining EF relations to vocabulary or engagement in joint attention likely fails to consider how differences in communication and language across contexts may relate to differences in cognition—and EF more specifically. Given the importance of communication and language within the early years of development, more work examining the relationship between language, communication, and early EF across different contexts (e.g., bilingualism, cultures) is likely needed.

3.4. The role of parents and family

Another sociocultural factor that has been considered frequently in early EF development is the role of parents and family. At this young age, the studies that have examined EF-family links have generally found at least one relationship. The most evaluated aspect is related to sensitivity, mostly assessed by observing a parent-infant interaction and rating sensitivity (e.g., appropriate and consistent responses to infant signals). The consensus across studies that examine parental sensitivity is that higher levels of sensitivity is related to better EF performance, although there were some qualifiers. Bernier and colleagues (2014) found more maternal sensitivity at 1 year of age related to better performance in conflict EF at 2 years of age, but only for infants who got more sleep at night. Zhang and Wang (2021) demonstrated high levels of maternal sensitivity at 6 months predicted better EF at 24 months, but only among children with low surgency (related to engagement with the environment and activity) measured at 6 months. Further, their work showed maternal sensitivity positively predicted EF in children with high looking duration to novel stimuli at 6 months (hypothesized to be related to less efficient information processing), but maternal sensitivity negatively predicted EF in children with low levels of looking duration to novel stimuli (hypothesized to be related to more efficient information processing). Ribner et al. (2022) showed that mother and father sensitivity uniquely positively predicted children’s WM, inhibition, and shifting at age 2. Although work by Li et al. (2022) showed maternal sensitivity at 14 months (but not 4 months) positively associated with infant inhibition concurrently, but paternal sensitivity was not a significant predictor of EF. Finally, Bernier et al. (2010) demonstrated that maternal sensitivity was positively related to several EF tasks at 18 and 26 months, but autonomy support was the strongest predictor when assessed with other maternal factors at each age.

Thus, there is also the possibility that factors other than maternal sensitivity may be strong predictors of children’s developing EF. For example, Ribner et al. (2022) also found maternal and paternal autonomy support uniquely predicted EF at age 2. Bernier et al. (2010) assessed maternal sensitivity, maternal mind-mindedness (assessed by coding how many times mothers commented on mental terms while talking to the child in a free play session) and autonomy support (also related to scaffolding in which parents offer and guide children with strategies appropriate their age to help them solve problems in addition to providing support while children make their own choices). They found the most support for autonomy support predicting EF at 18 and 26 months when all maternal factors were considered in the model. This aligns with other studies finding a link between maternal support and EF at 24 months (but not 18 months, Hughes et al., 2020). Marciszko et al. (2020) also found that parental scaffolding may play a role in EF, as their work demonstrated that infant social action understanding at 6 months (related to the formation of social internal working models predicting actor’s behavior) interacted with better parental scaffolding at 10 months in predicting simple inhibitory control 18 months. Others have found that parental expressed emotions relate to emerging EF, with more father criticism related to worse inhibition and father emotional over-involvement related to worse WM at 14 months of age. Better expressed quality of relationship for mothers related to better WM at 14 months (Blum & Ribner, 2022). Braren et al. (2021) also found that stronger mother-father cortisol linkages when the mother was in her last trimester (thought to be related to sensitivity to cues and capacity to adapt in a relationship) related to higher EF scores at 24 months. This was moderated by paternal average cortisol, demonstrating lower EF scores when fathers’ cortisol scores were elevated, and the mother-father cortisol linkage was weaker.

Taken together, the literature to date supports Ribner and colleagues’ (2022) conclusions that parents matter for the development of EF, but the influence of parents are likely multidimensional and varied. Their work demonstrates the potential for intergenerational transmission, with both mother and father EF relating to the development of children’s EF. Parenting practices also likely play a role in this transmission and contribute unique variance to EF development in toddlers. However, they also suggest that more work is needed, especially targeting additional measurement periods over the first 2 years to understand different mechanisms for intergenerational transmission over time. Further, it will be important for future work to integrate considerations of culture into the study of parenting and EF, as parenting shows important difference by culture and is an important means for transmitting values and norms within cultures (e.g., Bornstein, 2012)

4.0. Accounting for universality and context-specificity in early EF development

A comprehensive review of EF within the first two years of life may reveal two different but complementary lines of questioning that harken back to early debates in developmental psychology related to universality and context-specificity. As with most contemporary work in development, such sharp distinctions are unlikely. Rather, the question is how to strike a balance to understand two fundamental questions related to EF: 1) What elements of EF seem to be important to understanding a potentially universal domain-general ability in the conscious regulation of behavior (i.e., what is EF?) and 2) How does the sociocultural environment shape the development and potentially lead to different paths and execution for this potentially domain-general cognitive ability.

4.1. Accounts related to understanding universality in EF development

Many theoretical models of EF development approach EF as a universal domain-general regulatory ability focused on answering questions related to what EF looks like across development and understanding the underlying cognitive and neurological mechanisms related to progression in EF abilities. This developmental approach is valuable as a basis for understanding what EF is—important in the study of any psychological phenomenon. Although more work is clearly needed, the present review suggests several possibilities related to the structure and development of early EF. First, the earliest assessments of EF (approximately 12 and 18 months of age) are rarely correlated across tasks and the factor structures that fit best had tasks loading on their own factors. In addition, there are mixed results for longitudinal stability in the first two years, with earlier performance on EF tasks not always predicting EF performance at later ages—though stability does seem to improve with age. Finally, performance across age and cohesion between tasks generally improved with age (i.e., with better performance and strengthening reliability between assessments by approximately 24 months). Structurally, this may suggest a distinct period of development within the first two years of life when it is difficult to extract a common EF ability potentially because this is a period of emergence.

This algins with the unity/diversity framework proposed by Miyake and colleagues (e.g., Miyake & Friedman, 2012) in adults, suggesting that a common EF related to the ability to maintain task relevant information may be important to performance across all EF tasks and common to other EF abilities (e.g., updating, shifting). A single unitary factor has also been suggested to best fit EF performance across a battery of tasks in preschoolers (Wiebe et al., 2008, 2011). The fact that a unitary component of EF has not yet been well established in the first two years could mark this period of development as a period of EF emergence, when performance across EF tasks develop rapidly, begin to show more longitudinal stability, and show more coherence between task performance by 24 months. Related to Miyake and Friedman’s framework, this may be the period where a common EF ability related to maintaining task relevant information to guide behavior is emerging—and this common EF ability may begin to be applied across multiple EF tasks resulting in increased coherence between EF tasks.

There are also several theoretical perspectives that propose the first years of life as a period of emergence and rapid development. For example, Garon and colleagues’ integrative model of EF development (Garon et al., 2008) suggests that early prerequisite abilities in attention (e.g., selective attention) form a foundation for EF components to develop (e.g., children can hold simple representations early in the first year related to simple WM, but coordination with attention may allow for more complex WM abilities involving updating or manipulation to emerge in the second year). From this perspective, the lack of cohesion among EF tasks may be because basic skills in EF are emerging and developing before the age of 2 but are not yet integrated. They present the attention system and its integration with EF networks as a possible mechanisms for emerging and developing EF (e.g., Rothbart & Posner, 2001; Ruff & Rothbart, 1996).

Representational models of EF development (Marcovitch & Zelazo, 2009; Zelazo, 2004, 2015) would propose a similar pattern of development, although the underlying mechanisms driving the emergence of EF within the first few years of life focus more on the developing ability to form and reflect on task relevant representations to guide behavior. In these models, theorists may suggest that a lack of cohesion among EF tasks may be due to children’s developing ability to form representations that can be used to guide behavior in situations requiring cognitive control. For example, these frameworks have focused on how the development of language is important to the development of EF because this allows children to form and reflect on relevant information to guide behavior rather than habitual responses (e.g., labeling a hiding location in the A-not-B task may allow children to reflect on the novel B location rather than habitually search at location A, Miller & Marcovitch, 2011).

Importantly, both integrative attention based frameworks and representation frameworks would suggest that important neurological developments within the prefrontal cortex (Posner & Rothbart, 2013; Zelazo, 2015 see also Diamond, 2006) likely align with distinct patterns of EF development within the first two years. Further, both frameworks also broadly align with Miyake and Friedman’s (2012) conceptualization of a common EF. For instance, a primary factor in the emergence of EF from a representational standpoint is the ability to form and reflect on representations, which would be a necessary precursor to Miyake and Friedman’s conceptualization of a common EF (i.e., a representation of task relevant information must be formed and considered to guide behavior). Further, Garon et al.’s (2008) focus on the development of voluntary control of attention would be necessary to maintaining the task-relevant information once formed. Taken together, it is possible that critical developments in both representation and attention are linked and underlie EF (see Ruff & Rothbart, 1996, for discussion on how the development of language is important to attention development) and set the stage for the emergence of a common EF. However, clearly more work is needed to determine the source of the unique EF findings in the first two years and the developmental mechanisms underlying these findings.

4.2. Accounts related to understanding context specificity in early EF

The findings from the current review also suggests that while we may see some commonalities that suggest the first two years may be a distinct period of EF development, it is important to ground our understanding of EF in context. Most models of EF acknowledge that sociocultural factors (e.g., SES and parenting) impact the development of EF. However, as noted by Raver and Blair (2020), differences can sometimes be viewed from a deficit model suggesting discrepancies in performance on a construct like EF (that is typically studied and conceptualized with WEIRD - Western educated industrialized rich and democratic- samples as the prototype, Henrich et al., 2010) must be explained by understanding the cost or deficit related to the foundational theoretical model (e.g., poverty may incur a cost to EF by affecting foundational abilities in attention or biological development). The focus is less often on understanding the broader sociocultural context in which groups exist, potential inequities, and the strength and resilience communities exhibit in the face of these inequalities (Raver & Blair, 2020). Thus, to date, most theoretical models of EF development have not tackled how to fully consider the emergence, development, and execution of EF in context.

4.2.1. Integrating Context into Developmental Models of EF: A Social Representational Framework

One possibility would be to unite existing cognitive EF frameworks with sociocultural models of development. To this end, we explore a social representational framework inspired by two research traditions based in representational models of EF development described above and sociocultural perspectives of cognitive development. Figure 1 provides a depiction of this model, which is based on a competing systems model of EF represented by the black boxes and arrows (i.e., the Hierarchical Competing Systems Model, HCSM; Marcovitch & Zelazo, 2009). The HCSM proposes that goal directed behavior is driven by the interaction of two hierarchical foundational systems: the habit system automatically guides behavior based on habits or past experiences (e.g., immediately reaching for a desired toy) while the representational system guides behavior through conscious reflection on task relevant representations (e.g., recalling and maintaining the rule to wait your turn). The two systems compete to determine behavior, with reflection in the representational system able to override the habit system (both the possibility of reflection and for representations to override habits are represented by the large gray arrows in Figure 1, Marcovitch & Zelazo, 2009). The developing representational system is further described in the Levels of Consciousness Model (LoC; Zelazo, 2004), suggesting individuals can form and reflect on task relevant information at different levels of consciousness that develop with age. For instance, infants are primarily driven by unreflective and habitual responses (e.g., their awareness of a toy may be based in the habitual sensorimotor response to place stimuli in their mouth representing the toy as “something to suck on”), while toddlers who begin to use language may be aware of the same stimuli at a higher level (e.g., the toy may be linked to a word and semantic representation of rattle or represented as “something that you shake”). These developments in representation can help children control behavior (e.g., begin to control behavior with toys in different ways rather than putting everything in their mouth). These developments within representational abilities may help explain patterns in the first two years of EF, as this ability to use a higher LoC (e.g., labeling and language) occurs alongside the emerging coherence, rapid development, and improving stability of EF in the first few years. Although older children and adults can reach higher LoCs, it does not mean they are always operating at these high levels of representation and EF errors can occur when one fails to represent and reflect on the appropriate task relevant information to guide behavior.

Figure 1.

Depiction of the social representational model based on combinations of model depictions from the Hierarchical Competing Systems Model (Marcovitch & Zelazo, 2009) and the Iterative Reprocessing Model (Cunningham et al., 2007; Zelazo, 2015)

Researchers have extended this cognitive framework to social problem solving with the Iterative Reprocessing (IR) model (based on models like the HCSM and LoC model and represented in Figure 1 by the boxes and lines in blue). The IR model provides a more general theoretical model allowing for a more nuanced evaluation of behavior appreciating how experience may contribute to behavior across time. For instance, Cunningham et al. (2007) suggested the IR model could help explain many of our (potentially biased) behaviors in social situations. For example, when one encounters a stimulus (e.g., a person from an outgroup) potentially biased representations may be automatically activated (e.g., links to negative media related to the out-group). These initially biased representations can influence behavior (e.g., moving away from the target) and even affect the representations generated and processed about the target (e.g., interpreting an ambiguous behavior of the target as threatening). However, this model also allows for the consideration of context and experience (e.g., seeing the ambiguous behavior is harmless or positive) that can trigger further reflection and iteration on the stimulus and prompt the individual to consider different representations (e.g., positive outgroup representations) or suppression of negative representations. Therefore, the IR model provides a framework in which cognitive and social situations can be interpreted with an appreciation to the broader context and feedback from the environment.

Although there has been a call to place EF in context (Lewis & Carpendale, 2009; Raver & Blair, 2020; Sarma & Mariam Thomas, 2020) and rethink EF with considerations of the mental content (e.g., knowledge, beliefs, norms, values, and preferences) activated in goal-directed behavior (Doebel, 2020), there is less work exploring how representational models may help us address these sociocultural questions. A social representational framework suggests the dual-process model can be used to understand the control of behavior in many situations, however one needs to consider sociocultural factors that may shape our system for representation and the representations individuals brings to a situation requiring control (depicted in Figure 1 with red lines and arrows, see also Doebel, 2020; Sarma & Mariam Thomas, 2020). This is not to argue that the measurement and understanding of a domain-general ability in EF using latent analysis is not useful. On the contrary, the argument is that it is important to appreciate how the social and cultural environment may contribute to the emergence and development of a domain general EF (e.g., the development of language within a sociocultural context may transform our representational system), but also how the culture and environment may shape the representations and habits that operate within this developing domain-general ability (e.g., parents from cultures that focus on interdependence may stress resource sharing potentially leading to weaker habits of reaching for a desired toy and less of a need for EF in this particular sharing context).

4.2.2. Integrating a Social Representational Framework into the study of early EF

Importantly, this model suggests that sociocultural factors may shape the development of EF and mark the first two years as a critical period of EF emergence in several ways. For one, sociocultural factors impact the representational system we build (i.e., how we form, maintain, and update mental representations when control is needed). Social representational theories would suggest it is no coincidence that EF shows dramatic growth and increasing cohesion during the second year of life. This is the period when the foundation for the representational system is formed related to children’s ability to label objects within their environment (e.g., through language, e.g., Zelazo, 2004). This development is a communicative and social ability that emerges within our sociocultural context as humans and transforms the way we represent the world and control behavior (Luria, 1979; Tomasello, 1999; Vygotsky, 1986; Zelazo, 2004). Regarding the structure of EF, children may only begin to show cohesion between tasks when they develop the ability to form and reprocess representations across EF tasks—which is consistent with Miyake and Friedman’s (2012) conceptualization of common EF.

In many ways, this perspective is similar to the way other theoretical models may approach sociocultural influences on EF development and is still susceptible to a deficit-oriented framework (e.g., low-income backgrounds have been linked to lower verbal abilities possibly related to deficits in processing efficiency). However, although sociocultural factors may show links to foundational cognitive mechanisms with the potential to negatively impact EF, there is the opportunity to broaden the examination of why sociocultural elements may be related to differences (e.g., variation in linguistic representations related to SES could be due to availability, quality, and variety of resources in communities like library and literacy support, see Pace et al., 2017) and how the values, beliefs, and knowledge we form within our sociocultural environments shape the execution of EF within this model. A social-representational model would advocate for examining how environment may contribute to foundational differences in representational abilities underlying EF and its execution.

4.2.3. Future Directions

Future work is needed to examine the evidence for and feasibility of a framework like the social representational framework—and this work will likely involve novel research considerations and potential obstacles that come with broadening research participation and examination of samples that have traditionally been less studied (e.g., Henrich et al., 2010). It may involve collaborating with and involving members of the cultural communities in the research and employing mixed methods and open-ended questioning to better understand the values, beliefs, knowledge, and expectations surrounding EF and the situations in which children are asked to execute control (e.g., Miller et al., 2021; Raver & Blair, 2020). Approaching this type of study from a Social Representational framework may help us examine whether potential differences in values, beliefs, knowledge and expectations can help us understand EF differences across culture and backgrounds (e.g., (Keller et al., 2005; Lamm et al., 2018; Sabbagh et al., 2006; Tsethlikai, 2011; Wanless et al., 2013).

For example, consider the finding that parents of Chilean infants have reported higher levels of effortful control (i.e., a later-developing temperamental ability related to self-regulation thought to show connections with EF, Zhou et al., 2012) compared to US parents. Approached from a social representational framework, both Chilean and US children could be guided by the same struggle between the habit system (e.g., the prepotent response of obtaining a desired item) and the representational system (e.g., representing and reflecting on the waiting rule). However, the habits and potential for representation and reflection could be different because of their background. Children from Chile may have weaker habits related to the prepotent response of reaching for a desired treat based on their tendency to endorse high power distance values (i.e., unequal power distributions, vertical-based interactions, and rewards and sanctions based on rank, role, status, or age, Farkas & Vallotton, 2016). This may relate to children encountering negative consequences for failing to wait and children may even find it more habitual to orient to adult instruction. In addition, they may also have a stronger potential for representation and reflection when asked to wait based on their practice and positive reinforcement when waiting is achieved. This may provide an explanation within a cognitive framework for why we see cultural differences related to EF and may lead to several novel ways to test how context may operate within this representational framework. In addition, this framework would also predict that age would be an important factor to consider, as there may be fewer differences between cultures in younger children who have less accumulated experience with cultural value systems and a less developed representational system (e.g., Li et al., 2022).

Another potential future direction the social representational framework may focus on is measurement, suggesting that changing EF instruments to align with group values, expectations, and knowledge could relate to an increase performance in that group (e.g., examining EF in a task that prioritizes more autonomy and independence may relate to stronger EF performance in the US—in line with US values related to low power distance that value equal power distributions, symmetrical relations, and equitable rewards related to individual achievements). This may help provide a more critical approach to EF measurement that could integrate considerations of context and sociocultural influences when assessing EF performance. For example, an examination of the expectations and knowledge surrounding EF and language (which is hypothesized to be highly related to representational development and the emergence of a common EF) may align with Tsethlikai’s (2011) suggestion that a non-verbal WM task may be a better indicator of the cognitive development of American Indian children. This is based on her work demonstrating that children from the Tohono O’odham Nation exposed to the Tohono O’odham language may initially show lower scores on verbal cognitive measures, but the average scores on cognitive measures improve with age (consistent with other studies demonstrating a bilingual advantage) and the same deficit is not observed on non-verbal measures of WM. Further, Tsethlikai suggested results indicated that engagement in cultural activities may augment cognitive abilities. Thus, in lines with Raver and Blair (2020), this approach will rely on looking to families and communities to understand the knowledge they bring to different contexts involving EF and the importance of understanding the expectations and success communities have in addition to appreciating the resilience and strategies communities use in the face of obstacles.

Finally, it will be important moving forward for frameworks like the social representational framework to consider how the broader macrosystem related to economic, educational, and legal systems may influence EF. More specifically, researchers have called for the need for EF work to name and consider how inequalities that exist within our institutions affect EF and its development (Raver & Blair, 2020). More work is needed to integrate sociocultural considerations at this level into a social representational framework, but there may be potential moving forward. For example, it is possible that children facing discrimination or inequities may be managing additional (and potentially interfering) habits and representations that are activated during EF assessments. Take the example of stereotype threat, or the finding that adults and children underperform when a task-relevant stereotype related to their ingroup is made salient (Galdi et al., 2014; Steele & Aronson, 1995). It is likely that the “stimuli” presented at the start of an EF task in the social representational model involves more than just the competition between a habitual or prepotent response (grounded in one’s sociocultural experience) that can be overridden by one’s ability to represent task-relevant information that can be reflected upon to guide behavior. Other non-task relevant habits (e.g., implicit attitudes related to a task-relevant stereotype) and representations (e.g., explicit anxiety related to performing in line with the stereotype) may also be activated and influence individuals’ abilities to control behavior successfully (e.g., maintain focus on the task-relevant representation when competing representations are present). This, paired with less access to resources when faced with structural inequalities, would promote the consideration of structural and policy changes when looking for solutions in line with proposals by Raver and Blair (2020), such as a focus on decolonized or antibias curriculum and equity in funding within early education. It remains an open question as to how these larger sociocultural factors may affect EF development from a social representational model, especially during the first two years when children may be drawing heavily from their sociocultural environment while developing foundational representational abilities needed for EF.

5.0. Limitations and Conclusions

In the present paper we reviewed universality and context-specificity observed in early EF work within the first 2 years of life. The scoping review provides a useful approach to understanding early EF development—especially within the past decade, as roughly 80% (32 of 41) studies have been published within the last 10 years. Although not exhaustive, it also provides an overview of where we are currently in terms of understanding the structure of emerging EF across contexts. However, this work is not without its limitations. For one, it is important to note that this review only provides a snapshot of development (i.e., only studies with data from roughly the first 2 years of life). Many of the studies were part of larger longitudinal studies and expanding examination of early EF to how it relates to later EF development and other aspects of development (e.g., school readiness, social development) will be important. There are also other studies that include data from the first two years of life that could be informative (Espy et al., 1999; Garon et al., 2014; Mulder et al., 2014), but were not included because we could not isolate data from the age range of interests. In addition, data from the third year and older end of toddlerhood is likely also less studied and could be informative to understanding early EF (Howard & Melhuish, 2017; see also Carlson, 2005).

We present a social-representational framework as a possibility for considering both universal (i.e., through the role of representation in EF) and context specific (i.e., by examining how sociocultural factors shape the representational system and the representations one brings to a task) in the first 2 years of life. However, there are also limitations to this approach. Although the model suggests the importance of sociocultural factors in the development of the representational system and its content, more specification in the mechanisms related to its impact on EF would be useful. This is perhaps most notable in the understanding of how the sociocultural environment may shape the representational system, as this could occur at multiple levels. It may be useful moving forward to integrate larger systems models (e.g., Bronfenbrenner, 1979) or specific sociocultural mechanisms (e.g., Vygotsky, 1986) to better understand how sociocultural factors may influence EF through multiple pathways.

In addition, this framework is by no means the only approach to examine EF behavior in context—as systems models have argued for the examination of multiple interacting systems including ones that capture prior experience and the task environment (e.g., Spencer et al., 2001; Thelen et al., 2001). However, many theories (representational theories included) may focus more on understanding how constructs and processes in cognition operate on a more universal level. Although we know from developmental and individual differences work that these cognitive processes do not operate the same for all individuals, studies examining the sociocultural context are usually conducted in isolation from model testing and development. Thus, a clear picture of why and how we see this diversity in thought may be unclear. The social representational model provides one potential option for addressing some unanswered questions and recent calls to place the study of EF development in context.

Highlights.

• EF during the first two years may be a unique period of EF development with increasing coherence between task performance, significant EF development, and potential longitudinal stability.

• Developmental patterns for this age appear to hold across cultural and socioeconomic contexts.

• Evidence also suggests potential context-specific developments in EF during the first three years related to cultural context, socioeconomic status, language, and parenting.

• The social-representational model may provide a novel model for understanding both universal and context-specific developments in early EF.

Data Availability Statement

As this is a review article, data considered in the scoping review can be obtained from the original articles listed in Table 1.