Skip to content
Frontiers
Search

Original Research

Frontiers in Developmental Psychology ·

Volume 4 - 2026 | https://doi.org/10.3389/fdpys.2026.1881826

Using computerized measures of executive function skills in 3-year-olds—A cautionary tale from Kenya

  • Michael T. Willoughby1*corresponding author
  • Amanda C. Wylie1
  • Brent R. Collett2,3
  • Judy Adhiambo4,5
  • Collins Otieno6
  • Laura Mwangi5
  • Irine Oucho6
  • Perpetual Nyaguthi5
  • Priscah Lihanda6
  • Edemba Priscillah Wanini5
  • Ariana Magedson7
  • Susan Wamithi8
  • Beatrice C. Mutai4
  • Ruchi Tiwari7
  • Elizabeth Maleche-Obimbo4
  • Dalton Wamalwa4
  • Benson Singa6
  • Christine J. McGrath7
  • Sarah Benki-Nugent7,9
  1. 1Education Practice Area, RTI International, Research Triangle Park, Durham, NC, United States
  2. 2Center for Child Health, Behavior, and Development, Seattle Children's Research Institute, Seattle, WA, United States
  3. 3Department of Pediatrics/Developmental Medicine, University of Washington, Seattle, WA, United States
  4. 4Department of Pediatrics and Child Health, University of Nairobi, Nairobi, Kenya
  5. 5Kenyatta National Hospital, Nairobi, Kenya
  6. 6Center for Clinical Research, Kenya Medical Research Institute, Nairobi, Kenya
  7. 7Department of Global Health, University of Washington, Seattle, WA, United States
  8. 8Aga Khan University Medical College, Nairobi, Kenya
  9. 9Division of Allergy and Infectious Diseases, Department of Medicine, University of Washington, Seattle, WA, United States
Article metrics0 viewsView details
PDFCiteMetrics

Abstract

Computerized assessments of executive function (EF) skills are ubiquitous in research studies across the life span. This study reports on our experiences using a computerized assessment of EF skills in two studies that involved 3-year-olds in Kenya. Our initial attempt involved 64 children (M = 35.6, SD = 0.71 months old) who were participating in the Tunza Mwana study in rural Kenya. Our second attempt involved 161 children (M = 36.6, SD = 1.3 months old) who were participating in the ABC study in urban Kenya. Both studies administered a similar battery of tasks, using identical tablets, a common approach to assessor training, and included some cross-over in assessors. Most children were unable to pass the training items that are used to ensure they comprehend task instructions (rates of individual task completion were 17–70% in Tunza Mwana and 18–74% in ABC cohorts, respectively). On average, children in the Tunza Mwana cohort passed training items for 2 (M = 1.7, SD = 0.9) of 4 possible tasks, and children in the ABC cohort passed training items for 3 (M = 2.7, SD = 1.3) of 6 possible tasks. Among children who passed training items and proceeded to test items, task performance was typically low and often below chance levels. Household (i.e., access to electricity), caregiver (i.e., educational attainment), and child-level (i.e., observed attention, cognitive and language ability) characteristics were inconsistently associated with EF task completion and accuracy, which raised further questions about the validity of task scores. We discuss the results of researchers who are interested in tablet-based assessments of EF skills with 3-year-olds in low- and middle-income country contexts.

Keywords

1 Introduction

Performance-based assessments of executive function (EF) skills for use in early childhood have been available for over 30 years (e.g., Diamond, 1996; Espy et al., 2001; Zelazo et al., 1996). Whereas early research focused on EF for pediatric and high-risk populations, over time EF skills have been reconceptualized as a foundational aspect of normal cognitive development (Best and Miller, 2010; Hughes, 2011). EF skills have also increasingly become viewed as a common final pathway through which multiple risk factors undermine learning and academic success and increase risk for psychopathology (Blair and Raver, 2016; Freis et al., 2022; Hackman et al., 2015; Karalunas et al., 2021; Martel et al., 2017).

The first generation of performance-based EF assessments that were used in early childhood involved analog (i.e., hands-on, non-computerized) approaches that often included manipulable objects (e.g., card sorting) or social interactions (e.g., children make choices in response to instructions from an assessor or puppet). Carlson (2005) provided an influential summary of 22 analog tasks using a pooled sample of over 600 children ranging from 2-6-years of age. Her summary highlighted age-related differences in EF task difficulty and performance. For example, among the 17 tasks that were used with 3-year-olds, the probability of ‘passing' each task ranged from less than 10% to greater than 80%. These results provided practically useful information for researchers regarding task selection based on the age of their participants.

The second generation of EF assessments for use in early childhood involved computerized presentation of stimuli and recording of student responses (Howard and Melhuish, 2017; Wiebe et al., 2012; Willoughby and Blair, 2016). Compared to analog assessments, computerized assessments provided greater standardization (e.g., scripted instructions; fixed item presentation time; automated decisions about task discontinuation), more detailed sources of data (e.g., precise indicators of item accuracy and reaction time), and were more economical (e.g., eliminated data entry costs). Numerous studies have demonstrated that the use of computerized EF assessments with preschool-aged children exhibit strong psychometric properties, document developmental change, and are related to other aspects of early development (reviewed by Zelazo et al., 2016). The publication of the National Institutes of Health (NIH) Toolbox, which was promoted as appropriate for children as young as 3-years-old, legitimized and accelerated the use of computerized EF assessments in early childhood (Bauer and Zelazo, 2014; Zelazo et al., 2013).

Much of the initial support for the use of performance-based assessments of EF in early childhood derived from studies that involved analog tasks (Anderson and Reidy, 2012; Espy et al., 2006; Hughes and Ensor, 2011; Mulder et al., 2014; Willoughby et al., 2012; Wolfe and Bell, 2004). Moreover, analog tasks continue to be the most common approach for assessing EF in toddlers (McHarg et al., 2020; Morra et al., 2021; Schroer et al., 2023). Although subsequent studies established the feasibility and utility of computerized versions of EF tasks for use in early childhood (Howard and Melhuish, 2017; Muller et al., 2012; Willoughby and Blair, 2016), these studies often involved populations of varied ages (e.g., 3 to 6-years-olds), which may obscure a clear understanding of how well computerized assessments work with the youngest children. To be clear, many mixed-aged studies have not reported difficulties using computerized EF assessments with 3-year-olds (Clark et al., 2023; Kuhn et al., 2024; Wiebe et al., 2012). However, three recent studies involving the NIH Toolbox tasks are noteworthy. Becker et al. (2023) investigated the NIH Toolbox EF tasks in a large sample of exclusively 3-year-olds (N = 609; M = 39.6 months). Approximately 30% of children were unable to complete the practice items (and hence did not proceed to the test items). Among children who did complete practice items, 70–80% performed at or below chance levels on test items. Becker and colleagues suggested that the average 3-year-old may not have the requisite skills needed to complete the EF tasks in the NIH Toolbox. In a small mixed age (N = 89; M = 55.6, Range = 36–72 months) group of children who were living in emergency housing, Distefano et al. (2021) introduced developmental extensions (simplifications) of the NIH Toolbox EF tasks for children who failed practice items or an initial set of test items. In total, 45% and 70% of children required the developmentally extended versions of the flanker and dimensional change card sort tasks, respectively. Distefano and colleagues concluded that standard versions of NIH Toolbox tasks were too challenging for their population of students. In a large mixed age (N = 471; M = 55, Range = 36–75 months) group of children who attended an urban school district that was characterized by high rates of poverty and homelessness, Kalstabakken et al. (2021) also used a developmental extension of the NIH Toolbox EF tasks. In total, 62% and 58% of children required the developmentally extended versions of the flanker and dimensional change card sort tasks, respectively. These NIH Toolbox studies highlight that young children and/or children from disadvantaged backgrounds may have difficulty completing existing computerized EF assessments or that those assessments may require simplifications. These studies are consistent with the variable nature of EF skill development in early childhood (Garon et al., 2014; Theriault-Couture et al., 2025), as well as evidence that socioeconomic disadvantage undermines EF skill development (Blair and Raver, 2012; Noble and Giebler, 2020).

Most of what is known about the measurement of EF skills in early childhood, including the potential challenges of using computerized assessments with young and/or disadvantaged children, is based on studies that were conducted in high income country (HIC) contexts (Draper et al., 2023; Gaskins and Alcalá, 2023). Nonetheless, there is widespread interest in measuring EF skills in low-and-middle-income country (LMIC) contexts, where children are more likely to be exposed to risk factors such as infectious disease, environmental contaminants, or malnutrition that interfere with the development of EF skills (Segre et al., 2023; Ssemata et al., 2023; Truter et al., 2025). Some researchers have raised concerns that existing EF assessments may reflect cultural assumptions or implicit lived experiences that undermine their use in LMIC contexts (Alcalá, 2023; Gaskins and Alcalá, 2023; Jukes et al., 2024; Legare et al., 2018). Others have been more enthusiastic, highlighting positive associations of EF with learning and social experiences and negative associations with risk factors in comparable ways across HIC and LMICs (reviewed by Obradović and Willoughby, 2019).

McHenry and colleagues reviewed 16 tablet-based assessments of EF skills that have been developed and/or used in LMICs (McHenry et al., 2023). Four of the 16 assessments included multiple indicators of EF skills and were described as appropriate for use with children as young as 3-years-old. However, except for the Developmental assessment on an E-Platform (DEEP; Bhavnani et al., 2019; Mukherjee et al., 2020), the supporting studies involved children of mixed ages. For example, three previous studies described the adaptation process and general success of using Tangerine EF Touch tasks in rural and urban areas of Kenya (Willoughby et al., 2019a,b, 2021). These studies reported high rates of EF task completion and significant associations between EF task performance with processing speed, assessor-rated attention, and children's literacy, numeracy, and social-emotional skills. However, all three studies included mixed age study populations (3–6 and 4–7 years), which raise questions about how well these tasks work with the youngest or most vulnerable children.

The current study evaluated the feasibility and utility of using the Tangerine EF Touch assessment with 3-year-old children in rural and urban areas of Kenya. We capitalize on the results from two independent studies that included Tangerine EF Touch tasks as a part of their age 3-year visits. Neither study was designed to answer questions about which factors specifically contribute to 3-year-olds' performance on EF tasks. Rather, both studies used EF tasks to index individual differences in neurocognitive development among children with early life risk factors. The current study had three objectives. First, we described the proportion of 3-year-old children who were able to complete each EF task (i.e., passing training items was a precondition for task administration). Second, among children who completed each task, we described the proportion of test items that were answered correctly, with interest in whether average performance exceeded chance level responding. Third, we tested whether well-known correlates of EF skills, including household socioeconomic status and children's general cognitive development, were related to EF task completion and performance. If known correlates of EF were unrelated to task completion and/or performance, this would raise additional questions about the validity of EF task scores in 3-year-olds.

2 Methods

2.1 Overview

We report the results from two independent studies in Kenya that administered the Tangerine EF Touch battery at the age 3-year visit. Given study-specific interests and time constraints, the tasks that were administered varied slightly across studies. For maximal transparency, we present results for all tasks in each study.

2.2 Participants and procedures

2.2.1 Tunza mwana study

The Tunza Mwana study is a prospective cohort study conducted in Migori County, Kenya. Equal numbers of women who were living with and without HIV were enrolled during pregnancy, and mother-child dyads were followed to 36 months postpartum. The aim of the parent study was to understand how maternal HIV infection affects the association between human milk oligosaccharides, the infant gut microbiome, and child growth and neurodevelopment (Tiwari et al., 2025). Pregnant women, who were at least 18 years old at 28–42 weeks' gestation and who were planning to breastfeed for at least 6 months, were recruited during routine antenatal clinic visits from two regional hospitals that serve a predominantly rural population of families. In total, 350 mother-child dyads (50% of children were HIV-exposed and uninfected [CHEU]) were enrolled and followed from birth through 36-months of age.

The current study is limited to 64 children who attempted to complete the EF Touch assessment at the age 3-year visit (see Table 1; given poor performance, the EF assessment was discontinued for the remainder of participants). Participating children (50% male; 41% CHEU) were on average 35.6 months old at the time of the EF assessment (SD = 0.71; Range = 34–37 months); 16% had a World Health Organization height-for-age z-score below−2, consistent with stunting (World Health Organization, 2006). The EF Touch assessment was translated to Kiswahili and Dholuo; 56% of children were assessed in Kiswahili. Only 5% of children attended primary school or were in daycare or nursery. Mothers were on average 27.9 years old (SD = 6.2; Range = 18–40). About half of mothers had attained at least a secondary education (53%) and had access to electricity in the home (53%). Most mothers were married (80%). Household crowding averaged 0.71 rooms per person (SD = 0.43; Range = 0.14–1.00).

Table 1. Sample description and comparison.

Tunza Mwana (N = 64)ABC (N = 161)Compare
Continuous VariableNM (SD)NM (SD)t (df)
Age (Months)6435.56 (0.71)14836.63 (1.32)−7.60 (200.38)***
Bayley cognitive (raw score)63114.68 (9.73)154118.65 (7.19)−2.93 (90.84)**
Bayley communication (raw score)63123.19 (14.48)64127.61 (13.29)−2.17 (215)*
PSRA - attention/impulsive641.62 (0.82)1182.26 (0.40)−5.89 (79.94)***
Maternal age- years6427.91 (6.15)15827.28 (5.17)0.77 (220)
Rooms per person640.71 (0.43)1580.50 (0.23)78.11 (3.86)***
Categorical Variable*N**n* (%)*N**n* (%)χ2 (df)
Child sex- male6432 (50%)15872 (46%)0.36 (1)
HIV exposed6426 (41%)16016 (10%)28.15 (1)***
Stunted at age 3 years (HAZ < −2)6410 (16%)16022 (14%)0.72 (1)
Maternal edu-secondary or higher (ref: Primary)6434 (53%)158115 (73%)7.98 (1)**
Maternal marital status- married (ref: Unmarried)6451 (80%)158141 (89%)3.56 (1)
Has home access to electricity (ref: No)6434 (53%)158147 (93%)48.19 (1)***
Attends pre-primary school, daycare, or nursery643 (5%)14836 (24%)11.48 (1)***

*p < 0.05.

**p < 0.01.

***p < 0.001.

HAZ, height-for-age z-score using World Health Organization reference standards; PSRA, preschool self-regulation assessment; M, mean; SD, standard deviation.

2.2.2 ABC study

The Air Pollution Exposures in Early Life and Brain Development (ABC) Study is a prospective cohort study conducted in Nairobi, Kenya. The aim of the parent study was to investigate the association between exposure to air pollution and neurodevelopmental outcomes of children during their first three years of life (Benki-Nugent et al., 2025). Recruitment occurred in the Dandora health facility which was selected due to its proximity to the main dumpsite of the city of Nairobi, which spans approximately 50 hectares and where open waste burning is routine. Key inclusion criteria for enrollment included maternal age above 18 years, singleton pregnancy, and absence of any severe pregnancy complications. In total 400 women were enrolled at any stage of pregnancy, followed through delivery and mother-child dyads were follow up at the clinic up to child age of 36 months.

The current study is limited to 161 children who were age eligible and who attempted to complete the EF Touch assessment at the age 3-year visit (see Table 1). Children (46% male; 10% CHEU) were on average 36.6 months old at the time of the EF assessment (SD = 1.32; Range = 35–42 months); 14% were stunted. The EF Touch assessment was translated to English and Kiswahili; all but one assessment was completed in Kiswahili. Approximately one in 4 children received childcare outside of the home. Mothers were on average 27.28 years old (SD = 5.17; Range = 18–40). Most mothers had attained at least a secondary education (73%). A small proportion of mothers were unmarried (11%), lacked access to electricity in the home (7%), or lacked piped water (11%). Household crowding averaged 0.5 rooms per person (SD = 0.23; Range = 0.14–1.00).

2.2.3 Cohort comparisons

As summarized in Table 1, mother-child dyads in the Tunza Mwana and ABC cohorts did not differ with respect to maternal age (p = 0.44), marital status (p = 0.06), or child sex (p = 0.54). Although children in the Tunza Mwana cohort were slightly older than children in the ABC cohort (p < 0.001), the age range of children within each cohort was narrow. Compared to the Tunza Mwana families, children in the ABC cohort more likely to receive care outside of the home (p < 0.001). Moreover, the ABC families were more likely to have mothers with secondary education (p = 0.005) and were and were more likely to have access to electricity in the home (p < 0.001). However, ABC families had fewer rooms per person (higher crowding; p < 0.001) compared to Tunza Mwana families.

2.3 Measures

2.3.1 EF touch–overview

EF skills were measured using Tangerine EF Touch (Tangerine Central, LLC), a tablet-based battery of tasks that have been previously used with preschool-aged children in Kenya (Willoughby et al., 2019a,b). The psychometric properties of EF Touch tasks, including longitudinal measurement invariance, retest reliability, and criterion and predictive validity were previously established (Kuhn et al., 2024; Willoughby and Blair, 2016; Willoughby et al., 2019a). EF tasks were administered in a fixed order during a single assessment occasion. The fixed task order was used to reduce burden for assessors who were completing multiple assessment activities (TM task order: Training, Bubbles, Animal Go/No-Go, Mango-Banana, Pick the Picture, Animal Location. ABC task order: Training, Bubbles, Chicken-Dog, Mango-Banana, Pick the Picture, Silly Sounds Stroop, Something the Same, Animal Location). Tests were administered in English, Kiswahili, or Dholuo, based on the child's primary language. Assessments took an average of 23 min to complete (which was dependent on the number of tasks that a child completed). Because these tasks have been extensively described elsewhere, we only provide abbreviated task descriptions here. Exemplary screenshots of task stimuli and text translations are provided in Supplementary material.

The EF Touch battery begins with two tasks (Training, Bubbles) that orient children to the touch screen interface without making any EF demands. In Bubbles, children view blue circles (‘bubbles') that appear at random positions on the screen and are instructed to touch/pop the bubble as quickly as possible. Scores reflect the proportion of bubbles selected within time constraints and the mean reaction time of responses. Following the orientation tasks, children complete individual EF tasks that follow the same general structure. The assessor introduces the task using a standardized script and demonstrates a correct response. Next, children complete training items, which make minimal EF demands and are untimed. Training items (roughly 2–7 items depending on the task) are intended to help determine whether a child understands what is being asked of him/her (e.g., demonstrate a sequence of correct button presses for a go/no-go task in an untimed condition; correctly answer a working memory span item that included two elements). If a child did not answer training items correctly after up to two attempts (determined by the tablet—not the assessor), the task was automatically discontinued (and the task is marked as not completed). Children who correctly answer training items proceed to test items. Below, we describe the full set of tasks that were administered in the Tunza Mwana and/or ABC cohorts. In total, four EF tasks were administered in the Tunza Mwana cohort, and six EF tasks were administered in the ABC cohort. Three tasks overlapped across cohorts.

2.3.2 EF touch–inhibitory control

The battery included four tasks that measured inhibitory control: Animal Go/No-Go (a go/no-go task); Chicken-Dog (a combined go/no-go and continuous performance task), Mango/Banana (a spatial conflict task); and Silly Sounds (a Stroop-like task). All inhibitory control tasks had time constraints: children must respond within 3000–4500 milliseconds to items that involve overcoming a prepotent response. In Animal Go/No-Go, children are presented with a series of 40 animal drawings (i.e., cow, cat, dog, fish, bird, pig). Below each picture is a blue button. Children are instructed to touch the button as quickly as possible for every animal except cats (8 no-go items). Accuracy reflects the proportion of correctly answered (inhibited) pig items. In Mango/Banana, static drawings of a mango (bottom left) and banana (bottom right) remain on the screen, while 36 individual target drawings of mangos and bananas appeared, one at a time, at the top of the screen in either congruent (same side) or incongruent (opposite side) positions. Children were instructed to touch the static mango or banana drawing that matched the target drawing as quickly as possible. Accuracy reflects the proportion of correctly answered incongruent items. In Silly Sounds, children viewed a series of side-by-side line drawings of a cat and a dog (left and right animal placement varied across items), while audio clips of either a meow or woof sound played. In the first set of 17 items (congruent), children were instructed to touch the animal that makes the sound (congruent items). In the second set of 17 items (incongruent), children were instructed to be silly and to touch the animal that did not make the sound. Accuracy reflects the proportion of incongruent items answered correctly. In Chicken-Dog, a new task in the battery that was inspired by Clark et al. (2023), children are presented with a series of 60 drawings of either a chicken or dog. Children are instructed to touch the button as quickly as possible when they see the chicken but inhibit touching the button when they see a dog. The first items are used to capture inhibitory control, as chickens are more frequent than dogs (23 chickens, 7 dogs). Accuracy reflects the proportion of correctly answered (inhibited) dog items. The second 30 items are used to capture sustained attention, as dogs are more frequent than chickens (23 dogs, 7 chickens). Accuracy reflects the proportion of correctly answered chickens.

2.3.3 EF touch–working memory

The battery included two tasks that measured working memory: Pick the Picture (a self-ordered pointing task) and Animal Location (a visual-spatial Corsi-like task). In Pick the Picture, children viewed sets of 2, 3, or 4 pictures (two sets each) and were instructed to touch one of the pictures. The same set of pictures was then presented in a different order, and children were instructed to touch a picture that is not the same as what they have touched before. Accuracy reflects the proportion of correctly selected (non-duplicate) pictures after the first picture in a set (up to 12). In Animal Location, children view a 3x3 grid; sequentially, animals appear in cells of the grid at a consistent speed. Children were asked by the assessor to touch the cell (“box”) where individual animal appeared. In the first series of trials, animals were named by the assessor in the order that they appeared (forward accuracy); in later trials, animals were named by the assessor in the backwards order that they appeared (backward accuracy). Accuracy reflects the proportion of correctly answered items across respective trials.

2.3.4 EF touch–cognitive flexibility

The battery included one task that measured cognitive flexibility: Something's the Same (an item selection task). Children view two pictures that are described as matching with respect to their color, shape, or size (e.g., red chair and red flower match on color). They then view a third picture (e.g., blue chair) and are asked to identify which of the first two pictures matches the new picture in some way (e.g., red and blue chair match on shape). Accuracy reflects the proportion of correctly answered items out of 20.

2.3.5 Bayley scales of infant and toddler development, fourth edition (Bayley-4)

Cognitive, language, and motor development was assessed using a version of the Bayley-4 translated and adapted for use in Kenya (McHenry et al., 2021). Tests were administered in English, Kiswahili, or Dholuo, based on the child's primary language. Before testing study participants, examiners completed a training and certification protocol that included review of the Bayley-4 training video, in-person instruction and practice, and completion of video-recorded pilot assessments reviewed for quality control and inter-rater reliability. Examiners were required to complete at least one assessment with 90% item-level agreement and without significant administration errors. Throughout the study, 10% of the Bayley-4 assessments were video-recorded and reviewed for quality control purposes by the supervising psychologist to reduce drift. All Bayley-4 assessment reports were double-checked for errors including administrative, coding and calculation errors.

After each assessment, examiners indicated any concerns that could affect the validity of the assessment (e.g., child illness or fatigue, examiner error). Validity concerns were reviewed with the supervising psychologist, and one assessment was considered invalid. Any clinical concerns were reviewed with the study team and referrals were placed for clinical assessments when needed. The Bayley-4 was standardized in the United States, and the norm-referenced scores may not be appropriate for Kenyan children. As such, we present raw scores throughout.

2.3.6 Preschool self-regulation assessment assessor report

Following the EF Touch assessment, assessors completed the Preschool Self-Regulation Assessment (PSRA) Assessor Report, a 28-item questionnaire designed to capture children's attention, impulse control, activity level, sociability, energy, emotions, and aggression during the assessment (Smith-Donald et al., 2007). Except for three aggression items that were scored as binary indicators, all items were rated on a 4-point Likert scale. We focus on the Attention/Impulse Control Long scale, which takes a mean of 17 attention-related items and 2 aggression-related items. Higher scores reflect more advanced assessor report of attention and impulse control skills. Internal consistency for this scale was high (Cronbach's α = 0.95).

2.4 Analytic approach

We begin with a descriptive summary of the rates of EF Touch task completion (passing training items) and average task performance (among children who completed tasks) in TM and ABC samples. For tasks that were administered in both cohorts, we used chi square and t-tests to compare rates of task completion and performance. We used univariate logistic regression and Pearson correlations to test whether individual child, caregiver, or household factors were related to EF task completion and performance. We focused on univariate comparisons throughout because we were interested in whether any background characteristics were related to EF task completion and performance (not their unique contributions).

3 Results

3.1 Description of EF task completion and performance

3.1.1 Tunza mwana

On average, children completed 1.7 (SD = 0.9) of the four EF tasks that were administered. As summarized in Table 2, the individual rates of task completion were low for the Animal Go/No-Go (17%) and Animal Location (17%) tasks and moderate for the Mango-Banana (66%) and Pick the Picture (70%) tasks. Children who completed the Animal Go/No-go task answered an average of 48% of items correctly, which is indicative of chance responding (i.e., for each item, children make a dichotomous decision to touch a button based on for every animal unless it was a pig). Children who completed the Mango-Banana task answered an average 28% of items correctly, which was indicative of below chance responding (i.e., each item involves a dichotomous decision). Children who completed the Animal Location task answered an average of 21–22% of items correctly. Children who completed the Pick the Picture task completed 56% of items correctly. Consistent with the narrow age range, child age was not related to task completion (zero of four tests significant) and accuracy (one of five tests significant)—results are not presented.

Table 2. Rates of executive function task completion and accuracy.

Executive functionTunza Mwana (N = 64)ABC (N = 161)Compare
Task completionNn (%)Nn (%)χ2 (df)
Mango-Banana6442 (66%)158107 (68%)0.09 (1)
Pick the picture6445 (70%)152102 (67%)0.21 (1)
Animal location6411 (17%)14042 (30%)3.75 (1)
Animal Go/No-Go6311 (17%)–––
Silly Sounds stroop––15228 (18%)–
Something's the same––13339 (29%)–
Chicken-dog––160118 (74%)–
Task performance (Accuracy)*N*M (SD)*N*M (SD)*t* (df)
Bubbles (Proportion selected)630.86 (0.21)1610.95 (0.09)−3.30 (71.48)**
Bubbles (Mean RT in seconds)632.35 (0.61)1612.11 (0.52)2.96 (222)**
Mango-Banana420.28 (0.15)1070.29 (0.17)−0.48 (147)
Pick the picture450.56 (0.13)1020.55 (0.15)0.48 (145)
Animal location (Forward)110.21 (0.10)420.22 (0.11)−0.34 (51)
Animal location (Backward)110.22 (0.12)400.23 (0.13)−0.38 (49)
Animal Go/No-Go110.48 (0.27)–––
Silly sounds––280.33 (0.17)–
Something's the same––390.52 (0.11)–
Chicken-dog (No-Go; Inhibitory Control)––1170.66 (0.36)–
Chicken-dog (Go; Sustained Attention)––1180.71 (0.26)–
Total EF tasks– overlapping (Max = 3)641.53 (0.76)1591.58 (0.81)−0.40 (221)
Total EF tasks– all offered (Max = 4 for TM; Max = 6 for ABC)641.70 (0.90)1612.71 (1.26)–

*p < 0.05.

**p < 0.01.

***p < 0.001.

M, mean; SD, standard deviation; df, degrees of freedom (non-integer values are Satterthwaite method).

Task completed refers to the proportion of children who successfully passed training items after up to two attempts. Tunza Mwana task order: Training, Bubbles, Animal Go/No-Go, Mango-Banana, Pick the Picture, Animal Location. ABC task order: Training, Bubbles, Chicken-Dog, Mango-Banana, Pick the Picture, Silly Sounds Stroop, Something the Same, Animal Location.

3.1.2 ABC

On average, children completed 2.7 (SD = 1.3) of the six EF tasks that were administered. As summarized in Table 2, the individual rates of task completion were low for the Silly Sounds (18%), Something's the Same (29%), and Animal Location (30%) tasks and moderate for the Pick the Picture (67%), Mango-Banana (68%), and Chicken-Dog (74%) tasks. Children who completed the Silly Sounds task answered an average of 33% of items correctly. Children who completed the Something the Same task answered an average of 52% of items correctly. Children who completed the Mango-Banana task answered an average of 29% of items correctly. In each case, children's performance was indicative of at or below chance responding (i.e., for each item, children make a dichotomous decision). Children who completed Chicken-Dog answered an average of 66% (no-go, inhibitory control) and 71% (go, sustained attention) items correctly, which indicated above chance responding. Children who completed the Animal Location task answered an average of 22–23% of items correctly. Children who completed the Pick the Picture task completed 55% of items correctly. Consistent with the narrow age range, child age was not related to task completion (zero of six tests significant) or accuracy (one of eight tests significant)—results are not presented.

3.2 Predictors of EF task completion

We tested whether indicators of children's general cognitive development (i.e., Bayley cognitive and communication scores; assessor-rated attention) and family socioeconomic status (i.e., caregiver education; household access to electricity) were associated with the likelihood of completing each EF task. In the Tunza Mwana cohort, none of these predictors were significantly associated with the likelihood of completing any of the four EF tasks (see Table 3). In the ABC cohort, the Bayley Cognition score was associated with completion of two of the five EF tasks (i.e., children with higher Bayley Cognitive scores were more likely to complete the Animal Location [p < 0.001] and the Silly Sounds tasks [p = 0.04]; the Bayley Communication score was similarly associated with completion of the Animal Location task [p = 0.04]). The PSRA attention score was associated with completing the Chicken-Dog task (p = 0.04; i.e., children who were rated as having better attention were more likely to complete the task). In total, only 4 of 50 univariate tests were statistically significant, most of which would not survive adjustment for multiple testing. These largely null associations suggest that EF task completion was largely random and unrelated to other aspects of child cognitive development or socioeconomic status.

Table 3. EF task completion is weakly associated with household socioeconomic status and child cognitive ability.

PredictorMango-Banana OR (95% CI)Pick the picture OR (95% CI)Animal location OR (95% CI)AGNG OR (95% CI)Silly sounds OR (95% CI)Something same OR (95% CI)Chicken-dog OR (95% CI)
TMABCTMABCTMABCTMABCABCABC
Bayley cognitive1.83 (0.89, 3.76)0.81 (0.55, 1.19)1.35 (0.82, 2.21)0.97 (0.66, 1.44)2.25 (0.98, 5.18)2.32 (1.47, 3.66)***1.81 (0.82, 4.02)1.67 (1.01, 2.77)*1.36 (0.89, 2.06)1.24 (0.80, 1.93)
Bayley communication0.83 (0.45, 1.52)0.89 (0.61, 1.30)0.71 (0.33, 1.52)1.05 (0.74, 1.48)2.86 (0.84, 9.72)1.84 (1.02, 3.30)*3.48 (0.97, 12.44)2.05 (0.88, 4.74)1.73 (0.96, 3.11)1.20 (0.84, 1.70)
PSRA1.33 (0.87, 2.04)0.70 (0.36, 1.38)1.12 (0.72, 1.73)1.40 (0.66, 2.95)1.21 (0.71, 2.09)3.08 (1.00, 9.46)1.01 (0.60, 1.70)0.40 (0.12, 1.31)1.82 (0.62, 5.31)2.33 (1.20, 4.50)*
Maternal edu (Secondary)1.21 (0.43, 3.40)1.08 (0.51, 2.30)1.88 (0.64, 5.57)0.69 (0.31, 1.53)5.04 (0.99, 25.58)1.40 (0.59, 3.31)1.11 (0.30, 4.10)2.44 (0.71, 8.36)0.92 (0.40, 2.12)1.19 (0.54, 2.62)
Household electricity0.92 (0.33, 2.58)1.39 (0.37, 5.16)1.88 (0.64, 5.57)0.18 (0.02, 1.48)1.07 (0.29, 3.95)0.46 (0.13, 1.61)1.03 (0.28, 3.80)0.26 (0.04, 1.50)0.81 (0.19, 3.44)0.61 (0.13, 2.95)

*p < 0.05.

**p < 0.01.

***p < 0.001.

OR, odds ratio; CI, confidence interval; TM, Tunza Mwana cohort; ABC, ABC cohort.

Odds ratios represent univariate associations (no covariates); continuous variables are standardized (mean = 0, SD = 1).

3.3 Correlates of EF task performance (Accuracy)

Next, we tested whether indicators of children's general cognitive development (i.e., Bayley cognitive and communication scores; assessor-rated attention) and family socioeconomic status (i.e., caregiver education; household access to electricity) were associated with performance on individual EF tasks. Notably, many of these correlations were based on small sample sizes due to low rates of task completion.

3.3.1 Tunza Mwana

Children's performance on the Mango Banana task was significantly related to assessor-rated attention (r = 0.38, p = 0.01) and household electricity (r = −0.33, p = 0.03).Children who exhibited more attentive behaviors during testing performed better on Mango Banana. Counter-intuitively, children from homes with electricity performed worse on Mango Banana. Children's performance on Pick the Picture was significantly related to household electricity in the expected direction (r = 0.34, p = 0.02). None of the remaining bivariate correlations with individual task performance were significant (see Table 4). The Bayley Cognitive score was significantly associated with the total number of EF tasks that were completed (r = 0.38, p = 0.002). Although many correlations were based on a small number of observations (i.e., we had limited power to detect significant associations), most correlations were of small magnitude (|r| = 0.00 −0.20).

Table 4. EF task accuracy is weakly correlated with child cognitive ability.

CohortAccuracy scoresBayley- cognitive (r)Bayley-communication (r)PSRA attn.-impulse (r)Maternal edu (r)Household electricity (r)
Tunza MwanaMango-Banana a−0.080.040.38*−0.27−0.33*
Pick the picture a0.120.00−0.060.010.34*
Animal location forward b0.140.390.08−0.050.00
Animal Location Backward b−0.08−0.21−0.090.100.05
Animal Go/No-Go b−0.100.030.03−0.100.26
N tasks completed (all tasks, max of 4) c0.38**0.060.160.220.07
ABCMango-Banana d0.160.23*0.44***−0.050.03
Pick the picture d0.21*0.170.09−0.060.04
Animal location forward e0.250.210.20−0.020.10
Animal location backward e0.260.190.29−0.05−0.05
Silly sounds f0.280.40*0.39−0.14−0.02
Something's the same f0.090.180.190.45**0.09
Chicken-dog (Inhibitory control) accuracy g0.21*0.20*0.26*−0.120.06
Chicken-dog (Sustained Attn.) accuracy g0.150.20*0.46***−0.12−0.06
N tasks completed (all tasks, max of 6) h0.26**0.22**0.29**0.06−0.15

*p < 0.05.

**p < 0.01.

***p < 0.001.

RT, reaction time; PSRA, preschool self-regulation assessment.

aNs range from 42–45. bN = 11. c Ns range from 63–64. dNs range from 73–104. eNs range from 28–40. fNs range from 17–38. gNs range from 84–116. hNs range from 118–158.

3.3.2 ABC

Children's performance on the Mango Banana task was related to their Bayley Communication score (r = 0.23, p = 0.02) and assessor rated attention (r = 0.44, p < 0.001) in expected ways. Children's performance on the Pick the Picture task was related to their Bayley Cognitive score (r = 0.21, p = 0.04). Children's performance on the Silly Sounds task was related to Bayley Communication (r = 0.40, p = 0.04). Children's performance on Something's the Same was associated with maternal education (r = 0.45, p = 0.005). Children's performance on Chicken-Dog was related to their mean reaction time from the Bubbles task (r= −0.22, p = 0.02, for sustained attention score), Bayley Cognitive (r = 0.21, p = 0.03 for inhibitory control score), Bayley Communication (rs = 0.20, ps = 0.04 for inhibitory control and sustain attention scores), and assessor rated attention (r = 0.26, p = 0.02 inhibitory control score and r = 0.46, p < 0.001 for sustained attention score) in expected ways. Children's performance on the Animal Location was unrelated to all indicators (see Table 4). The Bayley Cognitive score (r = 0.26, p = 0.001), Bayley Communication score (r = 0.22, p = 0.005) and assessor rated attention score (r = 0.29, p = 0.001) were associated with the total number of EF tasks that were completed. Although many correlations were based on a small number of observations, most correlations were of small to moderate magnitude (|r| = 0.00 −0.25).

4 Discussion

There is global interest in assessing EF skills in young children. Here, we provide replicated evidence of the difficulties of using a computerized EF assessment in two, at-risk samples of 3-year-olds in Kenya. The rates of EF task completion (passing training items) were poor to moderate (17–70%). Children's general cognitive development and household socioeconomic status were largely not predictive of task completion. Among children who completed tasks (i.e., passed training items), task performance was often poor and consistent with chance-level responding. Although socioeconomic risk may not be a universal correlate of EF task performance (Cook et al., 2024), the inconsistent and weak associations between EF tasks and other indicators of cognitive development (Bayley, assessor-rated attention) raise concerns about the validity of these task scores. We consider the implications of these results for research and practice.

Across both cohorts, the rates of individual EF task completion were variable (17–70%) and average task performance was poor (21–56% accurate). These results differ markedly from three previous studies that used these tasks in Kenya (Willoughby et al., 2021; Willoughby et al., 2019a,b). In those studies, we observed higher rates of EF task completion (typically 80–90%) and average task performance that was not indicative of chance responding (60–80% accurate). The two most prominent differences between the current and those previous studies are children's age (most children in previous studies were 4–5 years old) and ascertainment method (previous studies included typically developing children who were recruited from pre-primary education settings). These results are consistent with recent studies that described challenges of using the NIH Toolbox EF tasks with 3-year-olds and with children from disadvantaged backgrounds (Becker et al., 2023; Distefano et al., 2021; Kalstabakken et al., 2021). The participants in our cohorts were young (3-years-old) and selected based on family risk factors (i.e., prenatal exposure to HIV in Tunza Mwana cohort; airborne exposure to toxicants in ABC cohort). The confounding of age, family risk, and setting undermines our ability to distinguish the relative importance of these factors on EF task performance.

The rates of task completion and task performance were remarkably similar for children in the Tunza Mwana and ABC cohorts. This was true despite greater socio-demographic risk factors (i.e., lower caregiver education, out-of-home care, and household electricity) and lower general cognitive development (i.e., performance on the Bayley and the Bubbles task; assessor rated attention) among children in the Tunza Mwana compared to ABC cohort. Despite the generally low rates of task completion, some tasks worked better than others. For example, approximately two-thirds of children in both cohorts passed the training items for Mango-Banana and Pick the Picture tasks (a similar result was evident for the Chicken-Dog task in the ABC cohort). These tasks may have been more successful because they had simple, explicit instructions and because the training items made few attentional or working memory demands. For example, in Mango-Banana, the training items require children to touch the mango [banana] button when they are presented with the mango [banana] picture (simple matching). In Chicken-Dog, the training items require children to touch the chicken but not the dog. Although Chicken-Dog and Animal Go/No-Go both use a go/no-go paradigm, Chicken-Dog uses two stimuli (chicken, dog) whereas Animal Go/No Go uses seven stimuli (multiple animal pictures). In Pick the Picture, the training items are limited to two pictures. Hence, children are only required to recall which of the two pictures that were presented was touched on the immediately prior turn. By comparison, tasks with lower rates of task completion involved more complex demands and/or more challenging language. For example, Animal Location required that children monitor the location (cell) of a series of pictures. Assessors reported that many children did not pay attention to the sequence of pictures (and the likelihood of guessing a correct response was low given six cells). For Silly Sounds Stroop, after children demonstrated that they associated barking and meowing sounds with dogs and cats, the training items require that they choose the opposite association (e.g., touch the dog when you hear a meow sound). It is not clear how many 3-year-olds understood the idea of choosing the opposite (incorrect) response, including the idea that this involves ‘being silly' (this is a language convention, not a translational issue). In addition to features of each task, test fatigue may have contributed to the low rates of task completion. In the ABC sample, the first three EF tasks had higher rates of task completion (67–74% completion) than did the last three tasks (18–30% completion). However, a similar pattern was not evident in the TM sample (17%, 66%, 70%, 17% completion for the four tasks, respectively).

Compared to EF Touch, fewer difficulties were encountered using the Bayley assessment. Among children who attempted EF Touch, only one Tunza Mwana participant had at least one Bayley subtest at 36 months that was determined to be invalid, whereas no Bayley subtests were considered invalid in the ABC cohort. Anecdotally, assessors reported that children were more engaged in the Bayley compared to the EF Touch assessment. We speculate that social engagement with assessors may have helped to sustain children's attention more so in the Bayley compared to EF Touch assessment. Children's familiarity and interest with the manipulatives that were used in the Bayley may have helped to maintain their interest better than their interactions with the tablet-based stimuli during EF Touch. Consistent with this speculation, we observed fewer issues with child refusal and test validity for Bayley tasks that involved toys and manipulatives vs. the stimulus book.

Multiple lines of work reinforce the idea that simplified task instructions and increasing child engagement and attention are essential for EF assessments with 3-year-old children in LMIC contexts. The DEEP assessment is one of the few tablet-based tasks that have shown some promise for use with 3-year-olds in a LMIC context (Bhavnani et al., 2019; Mukherjee et al., 2020). The DEEP assessment was iteratively developed to include features that engaged child attention (music, positive reinforcement, simple rather than complex images). Moreover, the tasks are presented in a game narrative, with the explicit goal of increasing children's engagement. Structurally, most of the DEEP tasks require children to complete simple actions (e.g., many tasks require children to make a single decision/action, like ‘popping a balloon' or dragging an object). Although not yet used in a LMIC context, Holmboe and colleagues have been iteratively developing the Early Childhood Inhibitory Touchscreen Task (ECITT), which measures response inhibition (a facet of inhibitory control) in children 18–36 months old (Holmboe et al., 2021). Like the Bayley, children can sit on a caregiver's lap during the assessment. They then engage with a tablet in front of them, including learning (through exploration) that screen touches produce feedback (animation, sound). Touches to one location (e.g., left side) are reinforced, which develops a prepotent (habitual) response. Once a habitual response bias has been created, the reward contingency switches to the other location (e.g., right side), which requires inhibitory control. Although the ECITT does not involve a narrative format, the ‘instructions' make minimal language demands, and children's responses are limited to simple button touches on the tablet. The ECCIT takes advantage of children's natural inclination toward exploration of the tablet and contingent feedback.

Jukes and colleagues have provided thoughtful guidance about a range of cultural factors that may influence the veracity of performance-based assessments of EF skills in LMIC contexts (Jukes et al., 2024). They noted that many EF tasks assume that children are motivated to comply with instructions from an unfamiliar adult, often require children to engage in arbitrary task rules, and may ask children to make fast and/or contrarian (‘silly', ‘wrong') responses. Children's exposure to many of these ideas varies across cultures and often first occur in the context of formal schooling. Notably, compulsory preprimary education begins at age 4 in Kenya. Children's participation in formal schooling may help to explain the large discrepancy in results from the current study with 3-year-olds compared to our previous studies in Kenya that involved children 4-years and older. We speculate that participation in pre-primary classrooms may serve to improve children's core knowledge (e.g., familiarity with and labels for colors, shapes, animals), help to improve their sustained attention, and increase their comfort level working with unfamiliar adults in an evaluation setting. These experiential factors may help to explain differences in using performance-based EF tasks with 3 compared to 4+ year olds.

In sum, our study serves as a cautionary note about using computerized EF assessments with 3-year-old children from disadvantaged backgrounds in Kenya. We incorrectly assumed that previous successes of using computerized EF assessments with preschool-aged samples in Kenya would generalize to the current cohorts. We speculate that differences in ascertainment methods (i.e., higher-risk vs. typically developing samples; community vs. school-based recruitment) and the age of participating children (i.e., exclusively 3-year-olds vs. 3-6-year-olds) explain differences observed here relative to previous studies. Applied researchers who work in LMIC contexts and who wish to measure EF skills in 3-year-olds should engage in extensive pilot testing of their tablet-based approach and focus on tasks that have minimal language and attentional demands of children. Alternatively, they may either explore analog alternatives (e.g., tasks that involve manipulable stimuli and/or that involve more social interactions with assessors) or focus on general indicators of cognitive development that have been developed for use with children under age 3 in LMIC settings (e.g., Global Scales for Early Development; McCray et al., 2023).

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fdpys.2026.1881826/full#supplementary-material

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Yvette Renee Harris, Miami University, United States

Reviewed by: Roisin Perry, University College London, United Kingdom

Mireya Chavez, Manuel Velasco Suárez National Institute of Neurology and Neurosurgery, Mexico

Statements

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This publication was made possible by primary funding from the National Institute of Environmental and Occupational Health (NIEHS), under award number R01ES032153 to S.B.N and funding from the National Institutes of Child Health and Human Development (NICHD) under award numbers R01HD096999 and P01HD107669.

Data availability

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

Ethics approval

The studies involving humans were approved by the Kenyatta National Hospital-University of Nairobi Ethics and Research Committee (P22/01/2021), the National Commission for Science, Technology and Innovation, and the UW Human Subjects Division (STUDY00012014)–all for ABC Study. The Tunza Mwana Study was approved by the Kenya Medical Research Institute Scientific and Ethics Review Unit (0140/3940), the National Commission for Science, Technology and Innovation, and the University of Washington Institutional Review Board (STUDY00007708). The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants' legal guardians/next of kin.

Author contributions

MW: Supervision, Methodology, Writing – original draft, Conceptualization, Writing – review & editing. AW: Methodology, Formal analysis, Writing – original draft, Data curation, Writing – review & editing. BC: Conceptualization, Investigation, Project administration, Supervision, Writing – review & editing. JA: Supervision, Writing – review & editing, Project administration, Investigation. CO: Validation, Writing – review & editing, Investigation. LM: Investigation, Project administration, Writing – review & editing. IO: Investigation, Validation, Writing – review & editing. PN: Investigation, Writing – review & editing, Project administration. PL: Writing – review & editing, Project administration, Supervision. EW: Investigation, Project administration, Writing – review & editing. AM: Software, Writing – review & editing, Project administration, Validation, Data curation. SW: Investigation, Writing – review & editing. BM: Writing – review & editing, Investigation. RT: Validation, Writing – review & editing, Methodology, Data curation. EM-O: Writing – review & editing, Investigation. DW: Writing – review & editing, Conceptualization, Supervision, Funding acquisition, Project administration, Investigation. BS: Conceptualization, Funding acquisition, Writing – review & editing, Investigation, Project administration, Supervision. CM: Investigation, Writing – original draft, Funding acquisition, Project administration, Supervision. SB-N: Investigation, Supervision, Funding acquisition, Writing – original draft, Conceptualization.

References

  1. Alcalá L. (2023). The developing of executive function skills through culturally organized autonomy and helping [Article]. Infant Child Dev., 32: e2460. doi: doi:10.1002/icd.2460
  2. Anderson P. J., Reidy N. (2012). Assessing executive function in preschoolers. Neuropsychol. Rev. 22, 345 – 360. doi: doi:10.1007/s11065-012-9220-3
  3. Bauer P. J., Zelazo P. D. (2014). The national institutes of health toolbox for the assessment of neurological and behavioral function: a tool for developmental science. Child Dev. Perspect. 8, 119 – 124. doi: doi:10.1111/cdep.12080
  4. Becker L., Condy E., Kaat A., Thurm A. (2023). How do 3-year-olds do on the NIH toolbox cognitive battery? Child Neuropsychol. 29, 521 – 542. doi: doi:10.1080/09297049.2022.2100337
  5. Benki-Nugent S., Were F. H., Riederer A. M., Gatari M., Karr C. J., Seto E. Y.. (2025). Air pollution exposures in early life and brain development in children (ABC): protocol for a pregnancy cohort study. BMJ Paediatr. Open 9: e002758. doi: doi:10.1136/bmjpo-2024-002758
  6. Best J. R., Miller P. H. (2010). A Developmental perspective on executive function. Child Dev. 81, 1641 – 1660. doi: doi:10.1111/j.1467-8624.2010.01499.x
  7. Bhavnani S., Mukherjee D., Dasgupta J., Verma D., Parameshwaran D., Divan G.. (2019). Development, feasibility and acceptability of a gamified cognitive DEvelopmental assessment on an E-Platform (DEEP) in rural Indian pre-schoolers—A pilot study. Glob. Health Action 12: 1548005. doi: doi:10.1080/16549716.2018.1548005
  8. Blair C., Raver C. C. (2012). Child development in the context of adversity experiential canalization of brain and behavior. Am. Psychol. 67, 309 – 318. doi: doi:10.1037/a0027493
  9. Blair C., Raver C. C. (2016). Poverty, stress, and brain development: new directions for prevention and intervention. Acad. Pediatr. 16, S30 – 36. doi: doi:10.1016/j.acap.2016.01.010
  10. Carlson S. A. (2005). Developmentally sensitive measures of executive function in preschool children. Dev. Neuropsychol. 28, 595 – 616. doi: doi:10.1207/s15326942dn2802_3
  11. Clark C. A. C., Cook K., Wang R., Rueschman M., Radcliffe J., Redline S.. (2023). Psychometric properties of a combined go/no-go and continuous performance task across childhood. Psychol Assess. 35: 353. doi: doi:10.1037/pas0001202
  12. Cook C. J., Howard S. J., Makaula H., Merkley R., Mshudulu M., Tshetu N., Scerif G., Draper C. E. (2024). Risk and protective factors for executive function in vulnerable South African preschool–age children. J. Cogn. 7: 58. doi: doi:10.5334/joc.377
  13. Diamond A. (1996). Evidence for the importance of dopamine for prefrontal cortex functions early in life. Philos. Trans. Royal Soc. B-Biol. Sci. 351, 1483 – 1493. doi: doi:10.1098/rstb.1996.0134
  14. Distefano R., Fiat A. E., Merrick J. S., Slotkin J., Zelazo P. D., Carlson S. M.. (2021). NIH Toolbox executive function measures with developmental extensions: Reliability and validity with preschoolers in emergency housing. Child Neuropsychol. 27, 709 – 717. doi: doi:10.1080/09297049.2021.1888905
  15. Draper C. E., Barnett L. M., Cook C. J., Cuartas J. A., Howard S. J., McCoy D. C.. (2023). Publishing child development research from around the world: an unfair playing field resulting in most of the world's child population under-represented in research [Editorial Material]. Infant Child Dev. 32: 13. doi: doi:10.1002/icd.2375
  16. Espy K. A., Bull R., Martin J., Stroup W. (2006). Measuring the development, of executive control with the shape school. Psychol. Assess. 18, 373 – 381. doi: doi:10.1037/1040-3590.18.4.373
  17. Espy K. A., Kaufmann P. M., Glisky M. L., McDiarmid M. (2001). New procedures to assess executive functions in preschool children. Clinical Neuropsychol. 15, 46 – 58. doi: doi:10.1076/clin.15.1.46.1908
  18. Freis S. M., Morrison C. L., Smolker H. R., Banich M. T., Kaiser R. H., Hewitt J. K.. (2022). Executive functions and impulsivity as transdiagnostic correlates of psychopathology in childhood: a behavioral genetic analysis. Front. Hum. Neurosci. 16: 863235. doi: doi:10.3389/fnhum.2022.863235
  19. Garon N., Smith I. M., Bryson S. E. (2014). A novel executive function battery for preschoolers: sensitivity to age differences. Child Neuropsychol. 20, 713 – 736. doi: doi:10.1080/09297049.2013.857650
  20. Gaskins S., Alcalá L. (2023). Studying executive function in culturally meaningful ways [Article]. J. Cogn. Dev. 24, 260 – 279. doi: doi:10.1080/15248372.2022.2160722
  21. Hackman D. A., Gallop R., Evans G. W., Farah M. J. (2015). Socioeconomic status and executive function: developmental trajectories and mediation. Dev. Sci. 18, 686 – 702. doi: doi:10.1111/desc.12246
  22. Holmboe K., Larkman C., de Klerk C., Simpson A., Bell M. A., Patton L.. (2021). The early childhood inhibitory touchscreen task: a new measure of response inhibition in toddlerhood and across the lifespan. PLoS ONE 16: e0260695. doi: doi:10.1371/journal.pone.0260695
  23. Howard S. J., Melhuish E. (2017). An early years toolbox for assessing early executive function, language, self-regulation, and social development: validity, reliability, and preliminary norms. J. Psychoeduc. Assess. 35, 255 – 275. doi: doi:10.1177/0734282916633009
  24. Hughes C. (2011). Changes and challenges in 20 years of research into the development of executive functions. Infant Child Dev. 20, 251 – 271. doi: doi:10.1002/icd.736
  25. Hughes C., Ensor R. (2011). Individual differences in growth in executive function across the transition to school predict externalizing and internalizing behaviors and self-perceived academic success at 6 years of age. J. Exp. Child Psychol. 108, 663 – 676. doi: doi:10.1016/j.jecp.2010.06.005
  26. Jukes M. C. H., Ahmed I., Baker S., Draper C. E., Howard S. J., McCoy D. C.. (2024). Principles for adapting assessments of executive function across cultural contexts. Brain Sci. 14: 318. doi: doi:10.3390/brainsci14040318
  27. Kalstabakken A. W., Desjardins C. D., Anderson J. E., Berghuis K. J., Hillyer C. K., Seiwert M. J.. (2021). Executive function measures in early childhood screening: concurrent and predictive validity. Early Child. Res. Q. 57, 144 – 155. doi: doi:10.1016/j.ecresq.2021.05.009
  28. Karalunas S. L., Antovich D., Goh P. K., Martel M. M., Tipsord J., Nousen E. K.. (2021). Longitudinal network model of the co-development of temperament, executive functioning, and psychopathology symptoms in youth with and without ADHD. Dev. Psychopathol. 33, 1803 – 1820. doi: doi:10.1017/S0954579421000900
  29. Kuhn L. J., Camerota M., Willoughby M. T., Blair C. (2024). A comparison of three executive function batteries in a preschool-aged sample. Children 11: 811. doi: doi:10.3390/children11070811
  30. Legare C. H., Dale M. T., Kim S. Y., Deak G. O. (2018). Cultural variation in cognitive flexibility reveals diversity in the development of executive functions. Sci. Rep. 8: 16326. doi: doi:10.1038/s41598-018-34756-2
  31. Martel M. M., Pan P. M., Hoffmann M. S., Gadelha A., do Rosario M. C., Mari J. J.. (2017). A general psychopathology factor (P factor) in children: structural model analysis and external validation through familial risk and child global executive function. J. Abnorm. Psychol. 126, 137 – 148. doi: doi:10.1037/abn0000205
  32. McCray G., McCoy D., Kariger P., Janus M., Black M. M., Chang S. M.. (2023). The creation of the global scales for early development (GSED) for children aged 0-3 years: combining subject matter expert judgements with big data. BMJ Glob Health 8: e009827. doi: doi:10.1136/bmjgh-2022-009827
  33. McHarg G., Ribner A. D., Devine R. T., Hughes C. (2020). Screen time and executive function in toddlerhood: a longitudinal study. Front. Psychol. 11: 570392. doi: doi:10.3389/fpsyg.2020.570392
  34. McHenry M. S., Mukherjee D., Bhavnani S., Kirolos A., Piper J. D., Crespo-Llado M. M.. (2023). The current landscape and future of tablet-based cognitive assessments for children in low-resourced settings. PLOS Digit Health 2: e0000196. doi: doi:10.1371/journal.pdig.0000196
  35. McHenry M. S., Oyungu E., Yang Z., Hines A. C., Ombitsa A. R., Vreeman R. C.. (2021). Cultural adaptation of the bayley scales of infant and toddler development, 3rd edition for use in kenyan children aged 18-36 months: a psychometric study. Res. Dev. Disabil. 110: 103837. doi: doi:10.1016/j.ridd.2020.103837
  36. Morra S., Gandolfi E., Panesi S., Prandelli L. (2021). A working memory span task for toddlers. Infant Behav. Dev. 63: 101550. doi: doi:10.1016/j.infbeh.2021.101550
  37. Mukherjee D., Bhavnani S., Swaminathan A., Verma D., Parameshwaran D., Divan G.. (2020). Proof of concept of a gamified developmental assessment on an e-platform (DEEP) Tool to measure cognitive development in rural indian preschool children. Front. Psychol. 11: 1202. doi: doi:10.3389/fpsyg.2020.01202
  38. Mulder H., Hoofs H., Verhagen J., van der Veen I., Leseman P. P. (2014). Psychometric properties and convergent and predictive validity of an executive function test battery for two-year-olds. Front. Psychol. 5, 1 – 17. doi: doi:10.3389/fpsyg.2014.00733
  39. Muller U., Kerns K. A., Konkin K. (2012). Test-retest reliability and practice effects of executive function tasks in preschool children. Clin. Neuropsychol. 26, 271 – 287. doi: doi:10.1080/13854046.2011.645558
  40. Noble K. G., Giebler M. A. (2020). The neuroscience of socioeconomic inequality. Curr Opi.n Behav. Sci. 36, 23 – 28. doi: doi:10.1016/j.cobeha.2020.05.007
  41. Obradović J., Willoughby M. T. (2019). Studying executive function skills in young children in low- and middle-income countries: progress and directions. Child Dev. Perspect. 13, 227 – 234. doi: doi:10.1111/cdep.12349
  42. Schroer L., Cooper R. P., Mareschal D. (2023). Assessing executive functions in free-roaming 2- to 3-year-olds. Front. Psychol. 14: 1210109. doi: doi:10.3389/fpsyg.2023.1210109
  43. Segre G., Cargnelutti C., Bersani C., Njogu W., Roberti E., Campi R.. (2023). Early detection of neurodevelopmental disorders in African children living in informal settlements in Nairobi. BMJ Paediatr. Open. 7: e002117. doi: doi:10.1136/bmjpo-2023-002117
  44. Smith-Donald R., Raver C. C., Hayes T., Richardson B. (2007). Preliminary construct and concurrent validity of the preschool self-regulation assessment (PSRA) for field-based research. Early Child. Res. Q. 22, 173 – 187. doi: doi:10.1016/j.ecresq.2007.01.002
  45. Ssemata A. S., Nakitende A. J., Kizito S., Thomas M. R., Islam S., Bangirana P.. (2023). Association of severe malaria with cognitive and behavioural outcomes in low- and middle-income countries: a meta-analysis and systematic review. Malar. J. 22: 227. doi: doi:10.1186/s12936-023-04653-9
  46. Theriault-Couture F., Matte-Gagne C., Bernier A. (2025). Child vocabulary and developmental growth in executive functions during toddlerhood. Dev. Sci. 28: e70010. doi: doi:10.1111/desc.70010
  47. Tiwari R., Singa B. O., Lihanda P., Diakhate M. M., Ochola E., Bunyige L.. (2025). Differences in growth trajectories in breastfed HIV-exposed uninfected and HIV-unexposed infants in Kenya: an observational cohort study. PLoS Med. 22: e1004781. doi: doi:10.1371/journal.pmed.1004781
  48. Truter B., Gillberg C., Slogrove A. L., Conradie P., Billstedt E., Thompson L. (2025). Neurodevelopmental problems in pre-school children in rural Western Cape, South Africa: is community screening feasible? Bmc Psychiat. 25: 348. doi: doi:10.1186/s12888-025-06791-7
  49. Wiebe S. A., Sheffield T. D., Espy K. A. (2012). Separating the Fish From the Sharks: a longitudinal study of preschool response inhibition. Child Dev., 83, 1245 – 1261. doi: doi:10.1111/j.1467-8624.2012.01765.x
  50. Willoughby M. T., Blair C. B. (2016). “Longitudinal measurement of executive function in preschoolers”, in Executive Function in Preschool Age Children: Integrating Measurement, Neurodevelopment and Translational Research, eds. J. Griffin, L. Freund, and P. McCardle (Washington, DC: American Psychological Association Press), 91 – 113. doi: doi:10.1037/14797-005
  51. Willoughby M. T., Piper B., King K. M., Nduku T., Henny C., Zimmermann S. (2021). Testing the efficacy of the Red-Light Purple-Light games in preprimary classrooms in Kenya. Front. Psychol. 12: 633049. doi: doi:10.3389/fpsyg.2021.633049
  52. Willoughby M. T., Piper B., Kwayumba D., McCune M. (2019a). Measuring executive function skills in young children in Kenya. Child Neuropsychol. 25, 425 – 444. doi: doi:10.1080/09297049.2018.1486395
  53. Willoughby M. T., Piper B., Oyanga A., Merseth King K. (2019b). Measuring executive function skills in young children in Kenya: associations with school readiness. Dev. Sci. 22: e12818. doi: doi:10.1111/desc.12818
  54. Willoughby M. T., Wirth R. J., Blair C. B., Family Life Project I. (2012). Executive function in early childhood: longitudinal measurement invariance and developmental change. Psychol. Assess. 24, 418 – 431. doi: doi:10.1037/a0025779
  55. Wolfe C. D., Bell M. A. (2004). Working memory and inhibitory control in early childhood: contributions from physiology, temperament, and language. Dev. Psychobiol. 44, 68 – 83. doi: doi:10.1002/dev.10152
  56. World Health Organization (2006). WHO Child Growth Standards: Length/Height-for-age, Weight-for-age, Weight-for-length, Weight-for-height and Body Mass Index-for-age: methods and development. World Health Organization.
  57. Zelazo P. D., Anderson J. E., Richler J., Wallner-Allen K., Beaumont J. L., Weintraub S. (2013). II. NIH toolbox cognition battery (CB): measuring executive function and attention. Monogr. Soc. Res. Child Dev. 78, 16 – 33. doi: doi:10.1111/mono.12032
  58. Zelazo P. D., Blair C. B., Willoughby M. T. (2016). Executive function: Implications for education (NCER 2017-2000). National Center for Education Research, Institute of Education Sciences, U.S. Department of Education.
  59. Zelazo P. D., Frye D., Rapus T. (1996). An age-related dissociation between knowing rules and using them. Cogn. Dev., 11, 37 – 63. doi: doi:10.1016/S0885-2014(96)90027-1

Citation

Willoughby, M. T., et al. (2026). Using computerized measures of executive function skills in 3-year-olds—A cautionary tale from Kenya. Frontiers in Developmental Psychology, Volume 4. https://doi.org/
Received
Accepted
Published
Volume
4 - 2026

Copyright

Copyright © 2026 Willoughby, Wylie, Collett, Adhiambo, Otieno, Mwangi, Oucho, Nyaguthi, Lihanda, Wanini, Magedson, Wamithi, Mutai, Tiwari, Maleche-Obimbo, Wamalwa, Singa, McGrath and Benki-Nugent.

License: https://creativecommons.org/licenses/by/4.0/

*Correspondence: Michael T. Willoughby

Article metrics

Views
0
Downloads
0
Citations
0
Shares
0

Related articles

PerspectiveFriluftsliv (open-air life) as a cultural site of emotion socialization in early childhood in NorwayEvi Petersen, Natália Meireles Santos da Costa, Tuva Beyer BrochReviewScreen exposure and cognitive outcomes in school-aged children and adolescents: a narrative review through the lens of Cognitive Load TheoryHiya Al Kalbani, Erin Kitt-LewisBrief Research ReportShort-term familiarity shapes social preference but not word learning in preschoolersJina Ahn, Erica A. Cartmill, Catherine M. Sandhofer