The search identified 277 studies (Fig. 1), of which 261 were excluded. Hence, only 16 studies that assessed basic EFs under 36 months old were considered (see Tables 2 and 3). Studies include diverse samples of children, such as those born premature or extremely premature, post-institutionalized, infants with ASD, bilingual, monolingual, and typically developing. Other studies included samples consisting of the mother-child dyad or the whole family. The sample’s ages range from 5 to 74 months old, although only findings in children younger than 36 months old were considered for this review.

Four topics describe the most relevant findings. The first explains the WM, inhibition, and CF tasks, as well as the two scales reported in the studies. The second addresses evidence suggesting that the unidimensional model can explain better all three basic EFs at these early ages. The third topic details the variables that may influence the measurement of the basic EFs. The last topic summarizes the findings of the variables that demonstrated an influence on the development of the EFs, to understand the complexity of measuring EFs at early ages.

Available Tasks and Scales

For this first topic, sixty-six tasks were examined from the 16 reviewed studies, some of which consisted of an adaptation of tasks described in other studies (see Table 2). Some tasks were assessed as composite scores (including the three basic EFs) and classified under different names (e.g., conflict tasks or cognitive executive function). Table 2 includes all this information organized by authors, type of EFs, task’s name including a citation of the study from which the task was obtained, and assessment ages. The following subsection describes the task features for each EF (i.e., WM, inhibition, and CF). Also, a battery and a questionnaire extracted from the literature are described in this review.

Working Memory Tasks

The WM tasks applied at earlier ages were two: hide the pots (or similar names) and hide and seek. These tasks were applied up to the age of 18 months (Garon et al., 2014; Treat et al., 2019), and both tasks required the child to find a toy under or behind some other object. The tools for hiding the toy may be different, but the goal (finding an object) is always the same. For example, a piece of fabric covers the toy for a waiting time (determined by the examiner) making it more complex to find the toy. As in most assessments, several practice trials are performed to proceed with the test. A score is obtained by counting the child’s mistakes.

Three studies used the spin-the-pots task, with a similar goal to hide the pots but with more pots and at older ages (Pozzetti et al., 2014; Treat et al., 2019). The task was applied with eight or more pots for children aged 28 months or older. Pozzetti et al. (2014) created an easier version using six pots for younger infants (< 24 months old). Another similar task is the three-box task that was applied by McHarg et al. (2020) for 14-month-old children. These tasks require placing several visually distinct containers on one tray. The examiner invites the child to place objects under the pots in a manner that left two pots empty. A piece of fabric is placed to cover the tray, and the tray is then rotated. Subsequently, the tray is uncovered, and the child is asked to choose one of the pots that contain an object. This is performed until the child finds a given number of objects or fails a maximum number of spins. The way of scoring differs among researchers: Some count errors and perseverations and others only whether or not the child achieves what is expected as a dichotomous measure.

Other tasks described in the reviewed studies were multi-location task, find the toy, multi-location multi-step, delayed alternation, six boxes, and visual attention. Nevertheless, its procedures and goals are similar to the tasks described above. Furthermore, studies revealed that performance on WM tasks improves with age.

Inhibition Tasks

To assess inhibition, tasks were identified in which the children are required to wait for a reward after a delay (inhibiting their behavior) or to inhibit a distracting stimulus to follow instructions correctly. The tasks in which the child was required to expect a reward after a delay were gift delay task, delay-of-gratification task, forbidden cookie, snack delay, delayed response, prohibition task, and self-control tasks. Basically, in these tasks, the children are offered a food or object that they like and are instructed to wait before giving it to them. In some studies, the child is asked to wait for a bell to ring, for the examiner to return, or for permission to take the object. As a quantifying assessment, the time the child can control the behavior is measured (Crivello et al., 2016; Treat et al., 2019). An alternative inhibition task, the statue task (Espy et al., 2004), also aims for the child to achieve control of a prohibited behavior (e.g., moving) but includes the interaction of the examiner. This consists of the children standing up and pretending to hold a flag in their hands for 75 s. Every 5 s, the examiner attempts to distract the child by recording any body movement of the child (e.g., eye opening or vocalization). The score is recorded in intervals. Two points are scored per error-free interval: one point if one error is made and zero if two or more errors are made.

The tasks shape Stroop task, children’s continuous performance task (C-CPT), and tricky-box task (Crivello et al., 2016; Espy et al., 2004) had a more complex goal, requiring the child to inhibit a distractor stimulus to achieve an instruction. The shape Stroop task consists of two phases: In the first phase (identification phase), the examiner introduces to the child three pictures with large fruits (i.e., apple, banana, and orange) and three pictures with the same fruits but smaller and placed below the large ones. Afterward, the small fruits are removed, and the children are asked to point to each fruit by telling the name and size, and feedback is given when they do it correctly. In the second phase (Stroop phase), the small pictures are placed below the large pictures but without matching them, and again, the child is asked to point to each fruit by telling the name and size; however, no feedback is given in this phase. The number of attempts to identify the small fruits in this Stroop phase is recorded. Regarding the C-CPT task, this is a computerized task in which the child is presented with images and sounds of several animals for 3 min. The sound of the animal should not match the image. After a familiarization phase, the child is instructed to press the button only when the animal (e.g., sheep) is visually presented, regardless of the sound it makes. Commission errors are scored, as they better measure poor modulation of response inhibition. For the tricky-box task, two identical but different-colored boxes are used. Each box has two red levers at the top and two transparent doors at the bottom. Behind the doors, there is a hidden toy. In one of the boxes, each lever lifts the door just below it. In the other box (conflicting), each lever lifted the opposite door. The above three tasks respond to a more complex inhibitory control.

Cognitive Flexibility Tasks

A commonly used task for assessing EFs at early ages is the A-not-B task, which is often used to assess the rudimentary form of CF. The search pattern necessary for this task develops from 5 months old in its simplest form and is more complex at 7 or 8 months old. A study with a sample of 10-month-old children allows differentiation of A-not-B responses in reaching and looking modalities (Johansson et al., 2014). The reaching response refers to when the child’s first response is to touch one of the presented screens. The looking response refers to the first look toward one of the screens for at least three frames (> 100 ms). This study supports that the looking modality in the A-not-B task is less cognitively demanding than the reaching modality. In the same context, Marcovitch et al. (2016) applied the A-not-B task to 5-month-old children; however, a limitation emerged as the A-not-B task could be inadequate for this age because the rule to stop the activity might be too strict. Hence, it might be more appropriate to measure object permanence and not CF at the age of 5 months old. The authors concluded that more research should be carried out considering recent studies on EFs for this age group and their relationship with other skills, such as language.

In addition to the A-not-B task, CF was assessed with tasks that require classification, such as the categorization task, the reverse categorization task, the DCCS scale, and shape and color sort (Pauen & Bechtel-Kuehne, 2016; Treat et al., 2019). These tasks have multiple procedures, depending on the child’s age. The simplest is when the child must sort toys according to some feature (e.g., size) into the appropriate pot. A more complex version requires children to sort toys in the opposite pot. Sorting can also be done by asking the child to select and sort specific cards (e.g., stars and trucks) among many different cards. The most complicated is to sort items by color and shape. The simplest version of this task was applied to children up to 17 months old, and the most complex was up to 24 months old. However, as in the previous EF measures, performance on these tasks improves with age.

Other tasks such as ball run task, flap book, spatial reversal, and spatial reversal with irrelevant color cues were applied to assess CF. However, the most relevant are those mentioned above.

Battery and Questionnaire to Assess EFs

Garon et al. (2014) developed a battery, which assesses WM, inhibition, and CF in pre-schoolers. They applied the battery to a sample of 261 children ranging from 18 to 67 months old. Authors showed that all EF measures were positively correlated with chronological and mental age. Moreover, the test was shown to be sensitive to age differences between 18 and 67 months old. Additionally, it was observed that developmental evolution was not the same among the three basic EFs. The factor structure of the three basic EFs resulted in two factors for each one; for WM, the factors were holding-in-mind and updating-in-mind. Regarding CF, the factors were task set and task switch. Finally, for inhibition, the factors were simple and complex inhibition. Reliability ranged from .70 to .93 in all tasks. The authors concluded that the early development of simple EFs (e.g., holding-in-mind, task set, and simple inhibition) is faster during the preschool stage and decelerates with age. Complex inhibition and CF task shift showed slight acceleration as age increases, while WM holding-in-mind shows deceleration. Therefore it seems that WM updating-in-mind exhibits a more gradual development than inhibition and CF.

The Early Executive Function Questionnaire (EEFQ) aimed to assess EF in children aged between 9 and 30 months old (Hendry & Holmboe, 2021). The exploratory factor analysis (EFA) and the confirmatory factor analysis (CFA) identified one factor for cognitive EFs with adequate psychometric properties which included the three basic EFs (WM, In, and CF). They called the factor “cognitive executive function” (CEF). Additionally, Cronbach’s alpha for CEF was .751. Authors studied regulation as another factor because it was theoretically important but did not obtain sufficient statistical support.

It is important to note that none of the selected studies in this contribution used BRIEF-P or Stanford Binet-5 to assess EFs; therefore, they were not described in this section. However, the usefulness of both tests to assess EFs from 24 months of life should be emphasized.

The Unidimensional Model of EFs

The factor structure most reported in the review responds to the unidimensional model, in which WM, inhibition, and CF constitute the common components of EFs. This common score has been referred to as conflict EFs (Bernier et al., 2012; Matte-Gagné & Bernier, 2011), composite score (Cuevas & Bell, 2014; Hostinar et al., 2012), or cognitive executive function (Hendry & Holmboe, 2021). Table 3 shows the results related to internal structure and reliability (when reported in the study), as well as other variables analyzed and the main findings of each study.

Garon et al. (2014) demonstrated that a common EF factor contains the three basic EFs, which are separated into two factors characterized by simple and complex tasks. These authors suggest that EFs before 36 months old may have different developmental trajectories that can be assessed with simple and complex tasks (Garon et al., 2014, 2018). Although, some studies showed 3 latent dimensions, separating WM, inhibition, and CF (McHarg et al., 2020; Pozzetti et al., 2014), the unidimensional model was more reported. Moreover, the unidimensional model has been obtained with an adequate sample size (200 or more participants) necessary for any AFE and CFA. Regarding reliability, many studies do not report it, and those that report internal consistency reliability showed a Cronbach’s alpha above .70, whereas others only indicate that it was low or limited. Nonetheless, interrater reliability was reported in five studies showing adequate values from .86 to .99. In summary, the unidimensional model appears to be consistent in studies of EFs before 36 months old. However, it should be considered that the factor structure and psychometric features should be reported in the studies to reinforce the support for this model.

Variables Related to EF Measurement

One of the challenges of measuring EFs is the difficulty of separating the measurement from other processes involved in EFs. In this review, attention, object permanence, and language were associated with EF measures.

First, there is a relationship between EFs and attention. Cuevas and Bell (2014) demonstrated that the type of attention used at 5 months old is potentially associated with EFs at older ages. The authors classified the participants according to attention style in short lookers (SL) and long lookers (LL) as reported by Colombo et al. (1991). They found that SL process information faster than LL since they encode global features, while LL tend to encode more localized features. Subsequently, they assessed the EFs of the same participants at 24, 36, and 48 months old, showing that SL scored higher on EFs throughout early childhood compared to LL, even when controlling verbal intelligence in the comparison. Likewise, Marcovitch et al. (2016) also included attention for 5-month-old children in two measures: (a) the percentage of looking time that was spent focusing on a puppet over the entire task and (b) the length of the longest look that was recorded. The most relevant of their results were as follows: (a) Attention is a predictor of the location of an object when it is hidden from the child’s view. (b) Maternal education is associated with the child’s maximum focus time on the object; for example, children of mothers who did not complete university or technical school present lower focus time. (c) Babies who hold attention for longer periods find the correct hiding place more often because the longer time of holding attention allows them to encode a greater amount of information, including the location of the hidden object. And, (d) at 7 months old, the child can relate the location of an object to its features; however, separate coding of features and location occurs at earlier ages.

Second, locating hidden objects is tied to the development of object permanence; however, when multiple locations are used to hide objects, the EFs are practiced (Marcovitch et al., 2016). This shows that difficulties in developing object permanence reflect deficiencies in inhibition and WM. For example, Pauen and Bechtel-Kuehne (2016) studied how tool use is associated with the development of EFs. They used Bechtel et al.’s (2013) tool selection task, which requires a transparent box with a tube and a ball (reward), as well as three sticks (tool) of different appearances and lengths. The sticks must be inserted through the tube, but only the longest one can reach the ball to drop it down a ramp and for the child to get it. First of all, the examiner inserts each tool into the apparatus in three consecutive attempts, from the shortest to the longest stick, showing the child that only the longest stick pushed the ball out of the tube (learning about the tool by observing others). The task continues with a training phase, which consists of several trials in which the child selects a tool for the examiner to insert into the tube and provides feedback to the child (learning about the tool based on feedback). Finally, there is a transfer phase in which the examiner replaces the training tools with new sticks and provides instruction to the child allowing 7 to 12 trials (transfer of tool knowledge to new situations). Their study found that children at 18, 20, 22, and 24 months old chose the correct tool after seeing another person apply it, increasing substantially the performance as the age group was older, especially between 20 (45%) and 22 months old (63%). Children who scored higher on WM made fewer perceptual errors after feedback; for example, those who waited longer before starting to eat the cookie in the inhibition task chose the wrong tool less frequently than those who did not wait; also, in CF, it was observed that switching skills helped to shift children’s perspective from paying attention on perceptually salient information toward paying attention on functionally relevant information. The results revealed that EFs are likely to impact the knowledge transfer of tools in young children.

Finally, language is an important variable related to EF development. Language skills operate as a support tool to inhibit impulsive responses, improve self-control, and have adequate social interaction (Matte-Gagné & Bernier, 2011). Several of the reviewed studies included a measure of verbal ability, language, or vocabulary to control for the influence on EFs (Bernier et al., 2012; Cuevas & Bell, 2014). The instructions of the EF tasks should also be adapted to the level of verbal development of children at this age.

Biological and Environmental Factors Related to EF Development

The review shows that basic EFs before 36 months old are influenced by variables such as prematurity, ASD, parenting, attachment, adverse early life experiences, bilingualism, and screen exposure. The findings about these variables are described below.

Premature birth is a relevant risk factor associated with EF difficulties and neurodevelopmental disorders. Pozzetti et al. (2014) studied this condition in children without significant brain damage and assessed the three basic EFs at 24 months old. The authors compared 72 premature infants and 73 as controls, finding that premature infants without significant brain damage have adequate cognitive development with difficulties only in CF which could be a precursor to deficits in young children. These difficulties could become more evident as they reach school age, when they are subjected to more complex tasks and the demands of the environment increase, resulting in academic and behavioral problems. Espy et al. (2004) considered a single sample that included typically developing children and premature children at 28 weeks gestation or more but without significant brain damage. They made this decision to increase the variability of performance by including children more likely to experience later mathematical difficulties. In their study, they determined that WM and inhibition could predict math skills later in development. However, the authors considered the limitation that the sample included children born preterm, as their findings might have differed with a more homogeneous sample.

Among the studies reviewed, only one included a sample of children with ASD in the age group of interest. Garon et al. (2018) applied their EF battery to 34 children with ASD and compared them with 255 children with typical development. Their results showed deficits in children with ASD in all the basic EFs assessed which were more evident when the tasks were appropriate for their level of development. These authors observed significant deficits in cognitive flexibility and inhibition rather than in WM, regardless of age. This implies that these ability deficits are not only observed in children with ASD but are also evident at early ages in the preschool period. The two abilities that most distinguished the groups were inhibition (simple and complex phase) and CF (simple and complex phase). In WM, the simple phase score had a slightly higher load than the complex phase, indicating that these deficits may be associated with developmental delays and not specifically with ASD symptomatology. Although deficits in inhibition were observed, CF was the one that best marked the severity of the symptoms in children with ASD. The study concluded that these measures could be candidates for endophenotypes, which would be an important line of research to further understand neurodevelopmental disorders.

The parent-child interaction and the quality of the environment at an early age also influence EFs. A favorable environment can have a positive impact, while an unfavorable one can have a negative structural and functional influence (Bernier et al., 2010; Bernier et al., 2012). Maternal care, specifically mind-mindedness, sensitivity, and autonomy support appear to influence conflict EF tasks, operating as a central mechanism in both cognitive and behavioral psychophysiological regulation, being maternal autonomy support the most important predictive factor of EFs at all ages (Bernier et al., 2010). Individual differences and aspects related to the family (e.g., stressful parenting or lack of opportunities) or the child’s verbal ability can affect the development of EFs (Matte-Gagné & Bernier, 2011). The longitudinal study of Bernier et al. (2012) looked at possible links between the quality of early care environment and EFs later in life and demonstrated that parenting and secure attachment were linked to conflict tasks at the age of three, even when taking into account socioeconomic status, child’s language, and previous performance on conflict tasks. Authors argued that the association between conflicting EFs and secure attachment is mainly attributable to two mechanisms: (1) the independent use of regulation strategies learned during emotionally evocative child-caregiver interactions and (2) more advanced psychobiological regulation, which supports the development of neural systems that underlie children’s ability to regulate thoughts and behaviors (Bernier et al., 2012). This is because care is presented in a harmonious environment, reducing negative emotional stimulation, which can affect physiological processes such as parasympathetic responses and cortisol reactivity. Parents’ early traumatic experiences could also impact the quality of care and as a consequence influence child EFs, specifically on lower performance in WM and inhibitory control scores due to harsh parenting attitudes (Treat et al., 2019).

Likewise, adverse experiences in a child’s first year of life can negatively influence child development. Hostinar et al. (2012) showed that post-adopted children with early life deprivation have lower performance in EFs, even after controlling for intelligence quotient (IQ) as a covariate. Specifically, the amount of time children spend in their biological family before being placed for adoption and the physical and social quality of the orphanage environment (reported by the parents and an expert from the adoption agency) influence the EF composite (WM, inhibition, and CF). These authors showed that even at the age of 2.5 years old, difficulties in EFs due to exposure to these experiences can manifest.

Proficiency in two languages may be a protective factor for EFs. Likewise, bilingualism may predict the development of EFs at later ages. Translation equivalents (TEs) are lexical representations of the same concept in different languages, which predict cognitive benefits in bilingual children. Increased TEs predicted stronger EF mechanisms, particularly in conflict tasks. In monolinguals, there was no relationship between increased vocabulary and conflict EF tasks, supporting that bilingualism produces this cognitive advantage. Bilinguals also showed superior selective attention and inhibitory control, because they must focus their attention on one language while ignoring the other (Crivello et al., 2016).

Finally, exposure to screens is associated with lower scores in inhibition (considering the response time to the prohibition task with a 5-s delay); this difference is not observed in WM or CF, suggesting a longitudinal association between exposure to screens and inhibition only. These results were observed even with other risk variables such as parental mental health problems or relationship difficulties (McHarg et al., 2020). Table 4 shows the variables that influence specific EFs (WM, inhibition, CF, or composite score).

Table 4 Variables influencing basic EFs