The neural correlates of inhibitory control in 10-month-old infants: a functional near-

infrared spectroscopy study

Abigail Fiske1*, Carina de Klerk2, Katie Y. K. Lui3, Liam Collins-Jones4, Alexandra Hendry1,

Isobel Greenhalgh5, Anna Hall1, Gaia Scerif1, Henrik Dvergsdal6, Karla Holmboe1

1 Department of Experimental Psychology, University of Oxford, United Kingdom

2 Department of Psychology, University of Essex, United Kingdom

3 Department of Psychology, University of Bath, United Kingdom

4 Department of Medical Physics & Biomedical Engineering, University College London,

United Kingdom

5 Child and Adolescent Mental Health Services, Oxford Health NHS Foundation Trust,

Oxford, United Kingdom

6 Nord University Business School, Department of Entrepreneurship, Innovation and

Organisation, Bodø, Norway

*Corresponding author: Abigail Fiske, Department of Experimental Psychology, University of Oxford, United Kingdom. Email: [email protected]

Acknowledgements

We would like to acknowledge Alison Jordan for her role in recruiting participants and organising test sessions, and Robert Cooper for his contribution to the custom probe design and code development used in this study. We also thank Jun Yin for sharing analysis scripts.

1

We offer our gratitude to the families who have continued to support and contribute to this project. This research was funded by the UK Medical Research Council (MR/N008626/1, PI:

Karla Holmboe) and by AF’s UK Medical Research Council Industrial Collaborative Awards in Science and Engineering (iCASE) studentship. The funder was not involved in the conceptualisation, design, data collection, analysis, decision to publish, or preparation of the manuscript.

Author Contributions

KH conceptualised the idea for the study and the original ECITT task design, and KH, CdK and HD designed the fNIRS version of the ECITT. AF, CdK and KH formulated the analysis plan. AF, CdK, GS and KH wrote the pre-registration relating to the individual differences analyses. KH, AF, AHa, IG, AHe and KL collected the data. AF and KL designed the video coding protocols and coded the videos. AF, KH and HD curated the data. AF and CdK processed and analysed the fNIRS data. AF, CdK and KH conducted the statistical analyses of the final data set. LCJ and AF visualised the data. HD programmed the ECITT software, and LCJ developed the pipeline to reconstruct the data and to plot/display the resulting reconstructed images. AF wrote the original draft. AF, KH, CdK, AHe, AHa, HD, and LCJ reviewed and edited the original draft. KH, CdK and GS supervised the study, and KH undertook the overall management and project administration of the larger project that this study was part of. KH acquired the funding for the study.

2

Abstract

Inhibitory control, a core executive function, emerges in infancy and develops rapidly across childhood. Methodological limitations have meant that studies investigating the neural correlates underlying inhibitory control in infancy are rare. Employing functional near- infrared spectroscopy alongside a novel touchscreen task that measures response inhibition, this study aimed to uncover the neural underpinnings of inhibitory control in 10-month-old infants (N = 135). We found that when inhibition is required, the right prefrontal and parietal cortices were more activated than when there is no inhibitory demand. Further, activation in right prefrontal areas was associated with individual differences in response inhibition performance. This demonstrates that inhibitory control in infants as young as 10 months of age is supported by similar brain areas as in older children and adults. With this study we have lowered the age-boundary for localising the neural substrates of response inhibition to the first year of life.

3

The early years of life represent a fundamental period in the development of executive functions (EFs), a set of core cognitive skills that facilitate the control of attention and behaviour in order to meet an adaptive goal. Executive function components such as inhibitory control, working memory and cognitive flexibility develop slowly across infancy and toddlerhood before improving rapidly in early childhood (Berg et al., 2020; Friedman et al., 2011; Garon et al., 2008, 2014; Howard et al., 2017; Wiebe et al., 2012; for reviews see,

Fiske & Holmboe, 2019 and Hendry et al., 2016). Accompanying these cognitive advances are important structural and functional maturation processes in the prefrontal cortex (PFC), a region of the brain consistently associated with EFs (Diamond, 2002; Fiske & Holmboe,

2019). Although there is some insight into the neural mechanisms involved in the development of working memory in infants and toddlers (Reyes et al., 2020; Wijeakumar et al., 2019), studies that identify the specific areas of the brain that are functionally associated with inhibitory control in infancy are rare. As such, knowledge about the neural substrates of

EF across this period of infancy and early childhood is limited.

From its emergence in infancy, inhibitory control is an essential EF that supports the deployment of cognitive control over goal irrelevant, non-adaptive behaviour (Friedman &

Miyake, 2017; Miyake et al., 2000; Miyake & Friedman, 2012). One important form of inhibitory control is response inhibition; the ability to inhibit a prepotent or well-learned response which is often motoric in nature (Friedman & Miyake, 2004). Whilst response inhibition abilities are immature during infancy, it is possible to reliably measure early forms of response inhibition from the second half of the first year of life (Hendry et al., 2021;

Holmboe et al., 2008, 2018, 2020). However, very little is known about the brain mechanisms supporting response inhibition in infancy and toddlerhood. The reason for this is twofold: a lack of age-appropriate tasks (Holmboe et al. 2020), and the difficulty in using neuroimaging techniques for measuring the awake infant brain in non-clinical settings (Lloyd-Fox et al.,

4

2010). In studies with older children and adults, canonical tasks such as the Stop-Signal

(Logan & Cowan, 1984) and the Go/No-Go task (e.g., Drewe, 1975) have frequently been used alongside functional magnetic resonance imaging (fMRI) to investigate the brain regions associated with response inhibition (Aron et al., 2007; Aron & Poldrack, 2006; Booth et al., 2003; Cope et al., 2020; Durston et al., 2002; Hampshire et al., 2010; Wager et al.,

2005). These well-established experimental paradigms provide a reliable measure of the neural correlates of response inhibition for older children and adults, but are unsuitable for the assessment of response inhibition skills as they emerge in infancy. This is because many of these tasks contain high working memory and language comprehension demands

(Holmboe et al., 2020) and neuroimaging modalities such as fMRI are not suitable for infants and young children.

Whilst there have been efforts to adapt or simplify EF tasks for use with toddlers (Bernier et al., 2010; Garon et al., 2014; Mulder et al., 2014) and young children (Berg et al., 2020;

Howard & Melhuish, 2017; Zelazo et al., 2013), there still exists a paucity of tasks that specifically assess response inhibition and the associated neural substrates in children under the age of two. Commendable efforts have been made in the EEG literature, whereby researchers have studied broad indices of neural activation (such as alpha power over frontal channels) to investigate inhibition-related brain activation in infancy. However, EEG has limited spatial resolution and, in contrast to gold-standard adult response inhibition paradigms, such as the Go/No-Go and Stop-signal tasks, these tasks are typically passive looking tasks that also place substantial demands on working memory (Bell & Fox, 1992;

Cuevas et al., 2012; Whedon et al., 2020).

The relatively recent addition of functional near-infrared spectroscopy (fNIRS) into the developmental researcher’s toolkit has made it possible to investigate neural activation in specific brain areas during the early stages of development (Fiske & Holmboe, 2019; Lloyd-

5

Fox et al., 2010; Wilcox & Biondi, 2015). fNIRS is a non-invasive, optical imaging technique that uses near-infrared light to measure the haemodynamic response; increases of oxygenated haemoglobin and decreases in deoxygenated haemoglobin concentrations in the cortex of the brain. Evidencing the feasibility and growing popularity of this technique in developmental research, growing numbers of studies are using fNIRS to investigate the neural correlates of

EF in preschool and school-age children (Buss et al., 2014; Kerr-German & Buss, 2020;

Monden et al., 2015; Moriguchi et al., 2018; Reyes et al., 2020; Wijeakumar et al., 2019;

Yanaoka et al., 2020). Studies using fNIRS to investigate response inhibition in this age group have found increased right lateral PFC activation when inhibition was successful

(Moriguchi and Shinohara, 2019), and that children and adults showed activation in the same areas (right PFC and parietal cortex) where inhibition was required, although children had a more immature response pattern (Mehnert et al., 2013).

This evidence is consistent with existing fMRI and transcranial magnetic stimulation (TMS) research showing that regions of the PFC and parietal lobe are activated during tasks of inhibitory control in both adults and children (Bunge et al., 2002; Cope et al., 2020; Durston et al., 2002, 2006; Tamm et al., 2002; van Belle et al., 2014). Specifically, the right dorsolateral prefrontal cortex (DLPFC; Casey et al., 1997; Cope et al., 2020; Durston et al.,

2006), the right inferior frontal cortex (as part of a fronto-basal-ganglia network; Aron et al.,

2004, 2014; Bari & Robbins, 2013; Chambers et al., 2009; Chikazoe, 2010), and regions within the right parietal cortex (Bunge et al., 2002; Chikazoe et al., 2009; Cope et al., 2020;

Durston et al., 2002; Gonzalez Alam et al., 2018; Mehnert et al., 2013; Osada et al., 2019) are reliably found to be active in tasks requiring response inhibition. Some studies have also found activation in the bilateral orbito-frontal when response inhibition is needed (Casey et al., 1997; Cope et al., 2020; Rubia et al., 2006; Tamm et al., 2002).

6

The current study

The overarching aim of this study is to address a fundamental gap in knowledge in that the functional neural mechanisms that underpin the emergence of response inhibition in infancy are currently unknown. By using fNIRS alongside an age-appropriate task designed specifically for infants and toddlers, this study aims to elucidate the neural underpinnings of response inhibition in 10-month-old infants. In doing so, we hope to uncover new insights into the neural mechanisms that support infants’ ability to exert control over their behaviour, as this skill is developing across the early years of life.

We used a version of the Early Childhood Inhibitory Touchscreen Task (ECITT; Holmboe et al., 2020), a novel measure of response inhibition that requires infants to inhibit their response to a prepotent (frequently-rewarded) location in order to instead make a response to an alternative location to receive the animated reward. The ECITT overcomes many methodological barriers by ensuring minimal language and working memory requirements, and including fun rewards that maximise task engagement in this population (Holmboe et al.,

2020), and has been shown to elicit inhibitory effects behaviourally in infants as young as 10 months of age (Hendry et al., 2021). Since a manual motoric response is required from infants, this more closely mirrors the structure and demands of canonical response inhibition tasks used with older participants in the fMRI/fNIRS literature.

The blocked version of the ECITT used in the current study consists of experimental and control blocks of trials, during which infants are required to make a manual response to one of two ‘buttons’ on the touchscreen in order to receive the reward. In control blocks, the target button (with a smiley on it) always appears in the prepotent location and as such there is no inhibitory demand. Experimental blocks contain trials where response inhibition is required (i.e., the smiley appears in the inhibitory/less frequently rewarded location), as well

7 as trials where the smiley appears in the prepotent location. Based on previous work (Hendry et al., 2021; Holmboe, et al., 2020), we predicted that response accuracy would be greater for prepotent trials, where there is no inhibitory demand, compared to inhibitory trials where response inhibition is required.

In this study we adopted a mixed exploratory and confirmatory approach. Due to the lack of existing fNIRS studies that use manual response-based tasks and the very limited prior evidence of neural activation during response inhibition in infant populations, we decided to not put forward any directional hypotheses about which specific brain regions would be associated with response inhibition at this age. However, based on findings in older children and adults, we designed our fNIRS probe to cover the bilateral PFC and parietal cortex. In our group-level analyses we first identified channels demonstrating a significant haemodynamic response and, within these, looked for channels in which the haemodynamic response was differentiated by block type (e.g., inhibitory-demanding blocks vs. blocks with no inhibitory requirement). We also identified the time course of the significant effect. Then, in a set of pre-registered analyses (https://osf.io/qs4h8) using these pre-identified channels and time bins, we investigated whether neural activation was associated with individual differences in response inhibition performance in 10-month-old infants.

Method Participants

Participants were 138 10-month-old infants (67 males) recruited for a longitudinal study on early executive function development. Three male infants were excluded as they did not meet the longitudinal study inclusion criteria. Data presented in this study were collected in the second of two visits to the Oxford Babylab, spaced approximately one week apart. In the first test session, infants completed a behavioural version of the ECITT, amongst other tasks.

8

Parents gave informed consent before participating with their child and received a £20 digital voucher and small Babylab branded gift for their participation. The longitudinal study received ethical approval from the University of Oxford Central University Research Ethics

Committee: R57972/RE010. See Supplementary Materials 1 (SM 1) for further details of the longitudinal study, including sample demographics (infant age, ethnicity and socioeconomic status indicators) and study inclusion criteria.

Apparatus fNIRS data were collected with the Gowerlabs Ltd. NTS continuous wave system at a sampling rate of 10Hz. Near-infrared light was emitted at 780 and 850nm to measure changes in concentration of deoxygenated (HHb) and oxygenated (HbO2) haemoglobin respectively.

The source levels were manually set, most commonly to 80% intensity (60% for the short separation channels), however this varied depending on the infant’s amount and type of hair.

The probe geometry consisted of 32 optical sensors with a source-detector distance of 25mm

(including two 12mm channels) that were fitted to an EasyCap. The probe was anchored onto three 10-5 reference points: FpZ, P1 and P2. A total of 46 channels covered the bilateral PFC and area around the intraparietal sulcus, see Figure 1 for a sensitivity map that indicates the approximate neural areas covered by our array. The headgear was fitted to the infant’s head so that the front of the cap sat just above the eyebrows, and the optode anchored to FpZ was positioned centrally between the eyebrows, see Figure 2. See also SM 2a for further cap placement information, and 2b for a channel map.

Figure 1 fNIRS Array Sensitivity Map

9

Note. This is a sensitivity heat map that was generated on a 12-month-old infant head model (Shi et al.,

2011) using the optode positions associated with this probe. These maps are scaled to the maximum sensitivity value in the grey matter mesh and are displayed on a log normalised scale. From left to right: frontal view, left view, superior view, inferior view and right view of the sensitivity of channels in this array.

Figure 2 fNIRS Cap Placement

Note. Consent for the publication of this image has been obtained from the participant’s parent via the Nature Human Behaviour research participant release form.

10

Early Childhood Inhibitory Touchscreen Task (ECITT)

In the ECITT, infants are presented with two blue ‘buttons’ on a touchscreen and are required to touch the button that contains the target stimulus (a smiley face icon) to receive an animated reward. The smiley occurs more frequently in one location (prepotent trials; 75%) than the other (inhibitory trials; 25%), as is common in response inhibition tasks such as the

Go/No-Go task (Casey et al., 1997; Durston et al., 2006). Therefore on inhibitory trials, infants must inhibit their response to the prepotent location and instead make a response to the other location to receive the reward. In the first test session, infants completed the original

(‘behavioural’) version of the task as described by Holmboe et al. (2020), with the exception that the task screen orientation was horizontal rather than vertical. Further details about the method as well as the results from the behavioural version of the task are reported in SM 5a, and test re-test analyses of task performance across the two versions are reported in SM 5b.

In the second session, participants completed a blocked version of the ECITT that was adapted for use with fNIRS, see Figure 3 for example trial sequences. This task was divided into control and experimental blocks (each consisting of six trials) that were separated by a neutral baseline video (moving abstract shapes accompanied by calm music). Since the task is response-contingent, the duration of each trial varied depending on how long the child took to respond. Baseline durations were jittered (12-17s) to minimise anticipatory neural activation ahead of active task blocks. In control blocks, the target always appeared in the prepotent location. In experimental blocks the target appeared on the prepotent side 50% of the time, randomised with the constraints that the first trial was always prepotent and the target cannot consecutively appear in the same location more than twice. Therefore, across the blocked version, the proportion of prepotent (75%) and inhibitory (25%) trials was the same as in the behavioural version. It is expected that prepotency in the control blocks would carry over into

11 experimental blocks, requiring response inhibition to be used on trials where the target was presented in the inhibitory location.

The blocked ECITT had no stopping point, but infants were encouraged to complete at least three blocks of each type to ensure enough reliable fNIRS data had been collected. The total number of blocks completed varied between infants. For those with valid fNIRS data (n =

59), the number of control blocks completed ranged from 3 – 7 (M = 4.85), the number of experimental blocks completed ranged from 3 – 7 (M = 4.58) and the total number of blocks completed ranged from 6 – 14 (M = 9.43), which is equivalent to 36 – 84 individual trials.

See section ‘Data Processing’ for details of validity criteria and data exclusions.

Figure 3

The Early Childhood Inhibitory Touchscreen Task

Note. Left: An example sequence of trials in a control block, where all trials were prepotent. Right: An example trial sequence in an experimental block. In both block types, a short animation is played when a correct response is made. Nothing happens when an incorrect response is made. A video of an infant completing the task is available here.

12

Consent for the publication of this video has been obtained from the participant’s parent via the Nature Human Behaviour research participant release form.

Procedure

Before the first test session, infants were allocated a prepotent side (left or right) that was counterbalanced between infants and remained the same for all test sessions. If infants made an incorrect response on the first prepotent trial of the ECITT in Session 1, the experimenter changed the prepotent side and re-started the task so that the target would appear in the other location. The participant would now be assigned to this new prepotent side, which then remained the same for all subsequent test sessions.

Infants were seated on their caregiver’s lap so that an experimenter could fit the fNIRS headgear whilst a second experimenter used bubbles and toys to distract the child. The parent and their infant were positioned at a small table, adjacent to the experimenter. The iPad

(presenting the ECITT) was held horizontally directly in front of the child and the experimenter ensured the iPad was within the infants’ reach. There were no demonstration or practice trials, however on the first trial, the experimenter cued the target by pointing and instructed the child to “touch the happy face”. This was done to ensure that the prepotency was established from the start of the task and was coded as invalid so that it was not included in analysis. Further verbal prompting and/or reward was given when necessary to increase engagement, but was kept to a minimum. After the first trial there was no further cueing of the specific location of the target. If the target location was cued by either the experimenter or the parent, the trial was coded as invalid (see section ‘Data Processing’ for details of validity criteria and data exclusions). During the animated rewards and baselines, the experimenter removed the iPad from the child’s reach (still within their sight) to prevent them from touching the screen and making an invalid (accidental or premature) response. Since this task

13 had no stopping point, the task was continued for as long as the infant was engaged with the task. The task was stopped if the infant became increasingly distracted or distressed, or at the request of the parent.

Data Processing

Behavioural Data Processing

Although the ECITT software automatically recorded the accuracy of responses, a detailed protocol that assessed the accuracy and validity of each trial was developed because infants’ responses were not always detected by the touchscreen. Videos of the testing sessions were coded offline by two coders (AF and KL) who gained excellent inter-coded reliability: κ =

.92 for accuracy, κ = .85 for validity across 21 videos (833 trials). Details of the coding scheme are described in SM 3a. Baseline blocks were coded as valid if the participant looked at the baseline video or other neutral location for at least 60% of the baseline duration; an excellent inter-coder reliability was established across 41 videos (1,715 coding incidences): κ

= .85. See SM 3b for further baseline coding scheme details.

fNIRS Data Processing

The NTS system was connected to a Window’s laptop, and a MATLAB script was used to send an exact time stamp (based on the internal laptop clock that was synced at least once a day) to the data stream recorded by the NTS. We used a custom MATLAB script to convert the universal output format (.txt) from the NTS system into .nirs format and to import event markers (start of each block) from the ECITT software, as well as the time sync information into the .nirs data. Discrepancies in time between that recorded by the ECITT software on the iPad, and the time shown on the laptop were minimal (M = 252ms, SD = 365ms, Range = 0 -

14

1556ms). Data were then processed in HomER2; a MATLAB based toolkit (Huppert,

Diamond, Francheschini & Boas, 2009). Invalid blocks and non-task related data were manually excluded (see SM 4a for details of data exclusion). Raw intensity data were converted to optical density and channels with very high or very low optical density (Range:

1e-03 - 1e+03) were excluded at the participant level. Participants with more than 2/3 of channels excluded were excluded from further analyses (see Data Exclusions below). Motion artifacts were identified and corrected using a spline interpolation followed by a wavelet transformation (Di Lorenzo et al., 2019). The motion-corrected optical density data were then converted to concentration changes in HbO2 and HHb using the modified Beer-Lambert Law with path length factors of 5.2 (HHb) and 4.8 (HbO2), see Scholkmann & Wolf (2013). A band-pass filter (high pass; 0.01, low pass; 0.80) was applied to remove low frequency noise and high frequency physiological signal from the data. See SM 4b for details of the parameters entered in this pre-processing stream in HomER2.

HbO2 and HHb concentration change data from each channel were block averaged over a period of 47 seconds, of which, two seconds contained data from baseline blocks. Baseline concentration change data was subtracted from the average haemoglobin concentrations in the 45 second experimental window. The baseline-corrected haemoglobin concentration data were divided into 9 time bins, each of a 5 second duration. Following investigations into the average block duration, we chose to focus our analysis on data from 0 – 35 seconds of the time course and as such, did not analyse data from the last two time bins (8 and 9). This is because due to the variable block duration (contingent on the duration of infants’ responses), these time bins captured baseline data from some participants and task data from others, rendering these time bins unsuitable for analysis. Results with these time bins included are reported in SM 8 and are in convergence with those reported in the ‘Results’ section below.

15

Data Exclusions

ECITT behavioural data were excluded from analyses if participants did not complete at least four blocks (two of each condition, with at least two valid inhibitory trials in experimental blocks; n = 13) or if an accuracy score of less than 60% on prepotent trials was obtained across all blocks (n = 1) or in experimental blocks (n = 0). See SM 6a for prepotent and inhibitory accuracy by block type (control and experimental). Following exclusions, a total of

121 infants had valid ECITT data. fNIRS data were excluded from processing and analysis if participants: did not complete the task (n = 2), did not wear the fNIRS cap during the task (n = 7), or if the data file was missing

(n = 1), or if the placement of the cap was not valid (n = 19). In order to ensure a sufficient number of trials, data were also excluded if participants did not complete at least three blocks of each condition in the ECITT task (n = 26), or if, following the manual removal of invalid blocks the participant no longer had at least three valid blocks (n = 13). Finally, data were excluded if more than 2/3 of channels were over- or under-saturated (n = 8). See SM 4a for further details on participant exclusions. The final sample of participants with valid fNIRS data was 59. This level of data attrition is similar to what is seen in other infant fNIRS studies

(e.g., Bulgarelli et al., 2019; Lloyd-Fox et al., 2019; Reyes et al., 2020), yet our sample of valid data is substantially larger than typical sample sizes for infant fNIRS studies (e.g., de

Klerk et al., 2019; Lloyd-Fox et al., 2015).

Behavioural Measures

A mean accuracy score for valid prepotent and inhibitory trials on the ECITT was generated following video coding. An ‘adjusted accuracy difference (adjusted AccD)’ variable was created as a measure of response inhibition using the following formula: (prepotent accuracy

16

– inhibitory accuracy)/prepotent accuracy. The larger the difference between the mean accuracy score on prepotent and inhibitory trials, the larger the adjusted AccD score. A smaller difference resulted in a smaller adjusted AccD score, which is indicative of better response inhibition. The adjusted AccD is a variation of the ‘AccD measure’ (prepotent accuracy – inhibitory accuracy) that was used by Holmboe et al. (2020) in their original study. We generated the adjusted AccD score to control for instances where infants performed relatively poorly on both trial types (i.e., the adjusted score took the infants’ level of performance on the prepotent trials into account).

Statistical Analysis

Behavioural Analyses

Group-level analyses of the ECITT data were conducted in SPSS version 27 with the three dependent variables; ‘prepotent accuracy’, ‘inhibitory accuracy’ and ‘adjusted accuracy difference’ (adjusted AccD). All variables were tested for the assumptions of parametric, and when these were violated, appropriate non-parametric equivalents were also used to test for convergence. Data distribution plots, details of assumption tests, and the results of the equivalent non-parametric tests are reported in SM 5 (Session 1) and SM 6b (Session 2).

The full sample of participants with valid data (n = 121) was split into participants with valid behavioural (ECITT) data only (n = 62) and those who additionally had valid fNIRS data (n =

59) in order to test for performance differences by trial type between the two sub-samples and to determine whether the fNIRS data are representative of the full sample and not a specific product of data that survived exclusions. This was tested using an independent t-test and a mixed analysis of variance (ANOVA). Our results confirmed that there were no significant performance differences between participants with and without valid fNIRS data.

17

We also performed paired t-tests to examine whether there were any significant changes in performance across the two test sessions, and Pearson’s correlation analyses were conducted to investigate test re-test reliability of the ECITT across sessions. Both tests were run with the adjusted AccD measure, as well as with the prepotent and inhibitory accuracy variables and are reported in SM 6d. The test re-test reliability of the full sample (also including those who did not have valid fNIRS data) is reported in SM 5b.

All statistical tests were two-sided and effect sizes (Cohen’s d and/or partial eta-squared) were calculated by SPSS when running the relevant statistical test.

fNIRS Analyses

For the purposes of this study, we define the haemodynamic response as a significant increase in HbO2 and/or decrease in HHb from baseline over time. Data from 42 channels (of

46) were included in statistical analyses at the group-level; two channels (Channel 11, right parietal cortex and Channel 15, left PFC) were excluded as less than 70% of participants contributed data to these channels, and data from the two short separation channels were also excluded from analyses. The short separation channels were included in our array with the intention to filter out superficial haemodynamic response (i.e., physiological noise) measured from the scalp and skull, however we did not implement short separation regression approaches due to the lack of a standardised approach to doing so within the infant fNIRS literature (Frijia et al., 2021).

We adopted a similar analysis approach to that used by de Klerk et al. (2018) and Lloyd-Fox et al. (2015). To identify channels demonstrating a significant increase in HbO2 and/or HHb decrease from baseline over time (main effect of time), separate repeated measures ANOVAs

(per channel) were conducted with time bin (7 levels) and condition (2 levels) as within-

18 subjects factors. In the analyses, a significant main effect of time would indicate significant haemodynamic concentration changes from baseline across the experimental window (35s).

For those channels demonstrating a significant haemodynamic response over time, further repeated measures analyses were conducted to investigate whether this response differed between experimental and control blocks, as indicated by a significant main effect of block type or a significant time × block type interaction. Greenhouse-Geisser corrected degrees of freedom and significance values were used because the channel-level data did not meet the sphericity assumption required for repeated measures ANOVA.

Individual Differences Analyses

In accordance with our pre-registration (https://osf.io/qs4h8), we conducted correlational analyses to investigate whether individual differences in neural activation were associated with individual performance differences in response inhibition. This included three confirmatory analyses (one-tailed) and one exploratory (two-tailed) analysis conducted in

SPSS version 27. Confidence intervals (CI; 95%) were calculated in SPSS on 1000 bootstrap samples. We analyse and report only the pre-registered channels and time bins, with the exception of one exploratory analysis that was conducted following the pre-registration (see

‘Results’). Since we were unaware of the distribution of the neural indices, we pre-registered that we would conduct both parametric (Pearson’s) and non-parametric (Spearman’s) correlational analyses and expected convergence in the results. Following pre-registration, all variables were tested for the assumptions of parametric. Details of assumption tests are reported in SM 7d.

An ‘inhibitory score’ variable (generated from the ECITT data and calculated as (1- adjusted

AccD)) was used as a measure of behavioural response inhibition, such that a larger

19 inhibitory score was indicative of better response inhibition ability. We reversed the adjusted

AccD score here to allow us to word our pre-registered predictions in terms of positive associations with brain activation in the experimental block (compared to the control block), rather than negative associations.

Channels that were identified in our group-level analyses as showing a significantly larger increase in HbO2 concentration and/or a significantly larger decrease in HHb concentration in experimental compared to control blocks were used as an index of neural activation. As such, we calculated the difference between HbO2 concentrations in experimental and control blocks

(experimental minus control), and the difference between HHb concentrations in control and experimental blocks (control minus experimental) and created an average of this difference measure across the identified channels or channel pairs. The specific time bins we studied

(for confirmatory tests, this was bins 3 – 5, 10 - 25 seconds of the block time course, and for exploratory tests, this was bins 4 – 6, equivalent to 15 – 30 seconds) were identified from our group-level analyses as demonstrating a significant difference in HbO2 and/or HHb between block types (control and experimental). See the pre-registration document for a full description of how these neural indices were calculated.

Head Modelling and Image Reconstruction

To allow us to visualise the data on an age-appropriate head template, a model of the infant head was produced from averaged structural MRI data of a 12-month-old cohort (Shi et al.,

2011). Group-level tissue masks were combined to produce a mask of the spatial distribution of the cerebral tissues (white matter, grey matter and cerebrospinal fluid). The inner skull border was delineated by the outside boundary of the cerebral tissue mask, while the scalp surface was defined using the Betsurf procedure (Jenkinson et al., 2005) where the group-level

20

T1-weighted MRI template was used as an input. All voxels situated between the inner skull border and the scalp surface were assigned to be extra-cerebral tissue; this is a combined label for scalp and skull, and is commonly done in infant head models due to the difficulty in discerning these two tissues in infant MRI data (Brigadoi et al., 2019; Collins-Jones et al.,

2021; Frijia et al., 2021). The resulting four-layer tissue mask (consisting of white matter, grey matter, cerebrospinal fluid and extra-cerebral tissue) was converted to a tetrahedral volume mesh and a grey matter surface mesh using the iso2mesh package ((Fang & Boas, 2009), see iso2mesh.sourceforge.net). The head model was scaled to the group mean head circumference measurement of the 59 infants in this study with useable fNIRS data. The positions of sources and detectors were registered virtually to the scalp surface of the head model using the Homer2 spring relaxation mechanism (Aasted et al., 2015). To model the transport of near-infrared light through the head model, TOAST++ ((Schweiger & Arridge, 2014), see http://toastplusplus.org) was employed to produce a forward model for each wavelength. Using the group-level block- averaged optical density data, the forward model was then used to reconstruct a time-series of images of HbO2 and HHb concentration changes for each condition. Image reconstruction was constrained to the grey matter nodes of the volume mesh, as per previous topographic approaches (Boas et al., 2004; Boas & Dale, 2005). The resulting reconstructed images were mapped to the grey matter surface mesh. Data preparation, meshing, forward modelling and reconstruction were facilitated by the DOT-HUB Toolbox (www.github.com/DOT-HUB).

Results Behavioural Results

Of the 121 participants with valid ECITT data, 59 also had valid fNIRS data (sub-sample A) and 62 participants only had valid behavioural data (sub-sample B). In both sub-samples, participants had higher mean accuracy scores on prepotent trials (sub-sample A; M = .928,

21

SD = .083, sub-sample B; M = .913, SD = .089) compared to inhibitory trials (sub-sample A;

M = .545, SD = .294, sub-sample B; M = .561, SD = .284). Results of an independent t-test revealed no significant difference in infants’ inhibitory performance (adjusted AccD) between sub-sample A (M = .402, SD = .332) and sub-sample B (M = .371, SD = .340); t

(119) = .497, p = .620, d = .336.

To test whether accuracy significantly differed by trial type or between sub-samples, we conducted a 2 x 2 mixed ANOVA with trial type (inhibitory or prepotent) as the within- subjects factor and sub-sample (A or B) as the between-subjects factor. Results revealed a significant effect of trial type; F (1, 119) = 161.411, p < .001, ηp2 = .576, confirming that infants are significantly more accurate on prepotent (M = .921, SE = .008) compared to inhibitory trials (M = .553, SE = .026). There was no significant effect of group: F (1, 119) =

.000, p = .994, ηp2 = .000, and no significant group by trial type interaction: F (1, 119) =

.278, p = .599, ηp2 = .002. This confirms that there are no significant behavioural performance differences between groups with and without valid neural data. Overall, these results suggests that the fNIRS data is representative of the full sample of infants and is unlikely to be a biased product of only those infants who tolerated the headgear or who have the sustained attention abilities needed to complete many trials.

fNIRS Results

Repeated measures ANOVAs identified 41 channels (out of 42 included in analyses) that demonstrated a main effect of time; a significant increase in HbO2 and/or decrease in HHb from baseline over time. A total of 32 channels showed a significant HbO2 increase over time, of which 27 also showed a significant HHb decrease from baseline over time, and six showed only a significant HbO2 increase. A total of eight channels showed only a significant

22

HHb decrease. The Benjamini-Hochberg procedure for controlling the false discovery rate

(FDR; Benjamini & Hochberg, 1995) was conducted (separately per haemoglobin signal) for

42 comparisons and all channels retained significance (except Channel 6, HbO2 only, which was then excluded from further analyses). See SM 7a for list of individual channels and test statistics.

From further repeated measures analyses with these 40 channels, we identified six channels that showed a significant main effect of block type (greater HbO2 increase/HHb decrease in experimental compared to control conditions), or a significant time × block type interaction, see Table 1. We found three channels in the right PFC (Channels 25, 26 and 32) and two in the right parietal cortex (Channels 10 and 12) that showed a significantly greater HbO2 increase in experimental compared to control blocks, and three channels in the right PFC

(Channels 25, 32 and 33) that showed a significantly greater decrease in HHb in experimental compared to control blocks. See Figure 4 for a t-statistic image of significant haemoglobin concentration differences between conditions, and Figure 5 for the haemodynamic response function plots for each channel. Similar t-statistic images of significant haemoglobin concentration change by block type (compared to baseline) are presented in SM 7a.

MATLAB versions of the t-stat figures are available on OSF (https://osf.io/mv47n/).

Channel Localisation

The cortical positions for these six channels showing significant effects were determined using the forward model (see Figure 5 for sensitivity profile of each channel). For each channel, the sensitivity values from the forward model mapped to the grey matter surface were used to compute a weighted average of grey matter node positions; the nearest grey matter node to the weighted average position was determined. Using the infant automated

23 anatomical labelling (AAL) atlas presented by Shi et al. (2011), the anatomical label of the nearest grey matter node was determined and was assigned as the cortical label of the channel. From this localisation process, pairs of significant channels that covered the same region of the brain were formed. This was done so that we could investigate the time course of the haemodynamic response in the channel pair associated with each region of the brain

(see section below for the results of the time course analysis).

Two pairs of channels showing significant HbO2 effects were formed: Channels 10 and 12, overlying the right parietal cortex, and Channels 25 and 26, covering the right PFC. Using the process described above, we were able to localise Channel 10 as covering the right inferior parietal gyrus and Channel 12 the right superior parietal gyrus. For analysis purposes, we will refer to this channel pair as the ‘right parietal cortex pair’. Channels 25 and 26 were localised to the right middle frontal gyrus, situated in the ventral region of the right DLPFC (Petrides

& Pandya, 2012), and as such this channel pair will be referred to as the ‘right DLPFC pair’.

For channels showing significant HHb effects (Channels 32 and 33), we formed a pair that was localised to the orbital middle frontal gyrus and as such will be referred to as the ‘orbital right PFC’.

For the purpose of the time course analysis, we decided not to include the HbO2 data from

Channel 32 in the frontal pair (with Channels 25 and 26), despite the fact that Channel 32 demonstrated a significant time × block type interaction for HbO2 (Table 1). This is because

Channel 32 is located in a more anterior part of the right PFC (specifically, the orbital PFC) whereas Channels 25 and 26 are located in the dorsolateral PFC. Therefore, we consider

Channel 32 independently for HbO2 and as part of the right orbital PFC pair for HHb. We justify this decision in full detail in SM 7b and in the pre-registration (https://osf.io/qs4h8/).

24

Table 1

Significant HbO2 and HHb Concentration Change Differences between Block Types

Channel Location Signal Statistic

* 2 10 Right inferior parietal cortex HbO2 Main effect of block type F (1, 51) = 6.374, p = .015 , ηp = .111

Time × block type interaction F (3.307, 168.653) = 3.258, p = .019*, ηp2 = .060

* 2 12 Right superior parietal cortex HbO2 Main effect of block type F (1, 50) = 5.113, p = .028 , ηp = .093

* 2 25 Right middle frontal gyrus/DLPFC HbO2 Main effect of block type F (1, 55) = 4.139, p = .047 , ηp = .070

Time × block type interaction F (3.915, 215.306) = 3.172, p = .015*, ηp2 = .055

HHb Main effect of block type F (1, 55) = 9.457, p = .003**, ηp2 = .147

* 2 26 Right middle frontal gyrus/DLPFC HbO2 Time × block type interaction F (3.907, 218.819) = 3.224, p = .014 , ηp = .054

** 2 32 Right orbital PFC HbO2 Time × block type interaction F (4.105, 229.852) = 3.874, p = .004 , ηp = .065

HHb Main effect of block type F (1, 56) = 8.041, p = .006**, ηp2 = .126

Time × block type interaction F (3.923, 219.713) = 5.170, p = .001***, ηp2 = .085a

33 Right orbital PFC HHb Time × block type interaction F (2.850, 165.322) = 3.910, p = .011*, ηp2 = .063

Note. * = significant at p < .05, ** = significant at p < .01 and *** = significant at p <.001, uncorrected for multiple comparisons. a remained significant when correcting for 35 comparisons; critical p-value = .0029. All other effects did not survive correction for multiple comparisons (HbO2, 32 comparisons; HHb, 35 comparisons; Benjamini & Hochberg, 1995).

25

Figure 4

Group-level T-statistic Images of the Contrast in Concentration Changes between Block Types

Note. Group-level T-statistic images of the contrast in concentration changes of each chromophore (HbO2 on the left and HHb on the right) between experimental and control blocks. Images are displayed in the space of a cortical surface derived from averaged structural MRI data of a 12-month-old cohort of infants

(Shi et al., 2011). Using this approach, all displayed T-statistic values are significantly different across block types (paired t-test) at the alpha level of p < .01. The HbO2 figure (left) covers the 10 – 25 second time frame, and the HHb figure (right) covers the 10 – 30 second time frame with which our analyses in significant channels were conducted. Since we found no significant HHb effects in parietal regions, the

HHb figure (right) only shows activation in frontal regions. A MATLAB version of this figure is available on OSF (https://osf.io/mv47n/).

26

Figure 5

Sensitivity Maps and Haemodynamic Response Function Plots for Channels Showing an

Effect of Block Type

27

Note. This figure demonstrates the sampling sensitivity (left) and haemodynamic response function (HRF) plots (right) of the six channels where a significant effect of block type or time × block type interaction was found. This is shown on an infant (12 months) averaged

MRI atlas (Shi et al., 2011). The HRF plots include the full 45 second time course. A grey line is present at the 35 second time point as this marks the end of our analysis window.

Data from 35 – 45 seconds were not included in analyses, see section ‘fNIRS Analyses’ in

Method. See also SM 2b for a channel map.

Time Course of the Block Type Effect

To investigate the time course of the block type effect, paired t-tests were conducted in each time bin on data from each pair of channels, and separately for Channel 25 (HHb) and

Channel 32 (HbO2). Results indicated that for the right parietal pair, there were three time bins in which there was a significantly greater increase in HbO2 concentration for experimental compared to control blocks; (10 –15s; p = .017, 15 –20s; p = .002, 20 –25s; p =

.001). For the pair of channels in the DLPFC, there were two time bins in which there was a

28 significantly greater increase in HbO2 concentration for experimental compared to control blocks; (10 –15s; p = .016, 15 – 20s; p = .045). However, for Channel 32 (right orbital PFC), there was a significantly (p = .021) greater HbO2 increase in control compared to experimental blocks from 0-5 seconds. Since the only significant effect occurred in a time bin where we would expect activation to still be building at the start of the block, and this difference was in the opposite direction to what was observed for the other channels, we did not further analyse or interpret the HbO2 data from Channel 32.

For Channel 25, there were five time bins in which there was a significantly greater HHb concentration decrease in experimental compared to control blocks (5 – 10s; p =.034, 10 –

15s; p = .005, 15 – 20s; p = .002, 20 – 25s; p = .001, 25 – 30s; p = .036). Similarly for the right orbital PFC pair (Channels 32 and 33), there were five time bins in which there was a significantly greater HHb concentration decrease in experimental compared to control blocks

(10 – 15s; p = .007, 15 – 20s; p = .008; 20 – 25s; p = .003, 25 – 30s; p = .004). See SM 7c for full results of the paired t-tests including effect size, and see also the HRF plots in Figure 5 for a visual illustration of the time course.

Association between fNIRS data and individual differences in ECITT performance

As stated in our pre-registration (https://osf.io/qs4h8), we conducted three confirmatory correlation analyses to investigate the association between neural activation and response inhibition performance. See sections ‘Indices’ and ‘Analysis Plan’ on the pre-registration for details of how the specific variables were calculated and the tests we performed.

Results of the one-tailed Pearson’s correlational test revealed that there was no significant association between individual differences in infants’ inhibitory score and HbO2 activation in the right lateral PFC pair; r (57) = -.056, p = .338, [CI = -.287, .156], or in the right parietal

29 pair; r (53) = .171, p = .106, [CI = -.134, .423]. These results were in convergence with the

Spearman’s rho correlational analysis; respectively, rs (57) = -.070, p = .298, [CI = -.313,

.158], and rs (53) = .071, p = .304, [CI = .232, .336]. This suggests that there are no significant associations between infants’ response inhibition performance and HbO2 concentration in the channels that cover these areas of the brain.

A modest, significant positive association was found between HHb concentration difference from 10 – 25 seconds in Channel 25 (right PFC) and inhibitory performance; r (54) = .255, p

= .029, [CI = .016, .446], see Figure 6. This suggests that the HHb concentration difference in

Channel 25 across this time window is weakly predictive of individual differences in infants’ response inhibition performance on the ECITT task. However, the results of the Spearman’s correlation did not meet the significance threshold; rs (54) = .216, p = .055, [CI = -.050,

.444]. Since the effect size for both correlation techniques are similar, it is possible that the non-significant result is due to the lower statistical power of the non-parametric test.

However, as the HHb difference variable for Channel 25 was normally distributed and the skew for the inhibitory variable was within an acceptable level, we are cautiously considering the Pearson’s correlation to be a fair indication of the true correlation. However, the significant result obtained from the Pearson’s correlation test did not survive correction for three comparisons (Benjamini & Hochberg, 1995) and so the possibility of this result being a false positive should be considered.

30

Figure 6 Association between the HHb difference score (Channel 25, Bins 3 – 5) and Inhibitory Score

Note. A larger inhibitory score is indicative of better response inhibition performance on the

ECITT. A larger HHb difference score reflects a larger HHb decrease in experimental compared to control blocks. The HHb concentration difference was calculated as: average baseline-corrected HHb concentration in control blocks minus average baseline-corrected

HHb concentration in experimental blocks. This was averaged across time bins 3 – 5 (10 – 25 seconds of the block time course) for Channel 25.

Exploratory Analyses

We also pre-registered that we would perform exploratory (two-tailed) correlational analyses to examine whether the difference in HHb concentration across time bins 4 – 6 (15 – 30 seconds of the block time course) in the right anterior PFC channel pair (Channels 32 and 33) was associated with infants’ response inhibition performance. Note that the broad label of

‘right anterior PFC’ was used in the pre-registration in reference to Channels 32 and 33, however these channels were later localised more specifically to the right orbital PFC. Our

31 justifications for running this as an exploratory test are outlined in the ‘Exploratory analysis’ section of the pre-registration document. We pre-registered that this association would be tested using both Pearson’s and Spearman’s correlational methods, since we did not know the distribution of the neural index prior to registration. Both variables were tested for the assumptions of parametric, and when these were violated, appropriate non-parametric equivalents were also used to test for convergence (reported in SM 7e).

Results of the Pearson’s correlation indicated that there was no significant association between HHb concentration difference in the orbital PFC and inhibitory performance; r (57)

= -.222, p = .092, [CI = -.423, -.043], whereas Spearman’s correlation indicated that a significant weak negative association was present; rs (57) = -.281, p = .031, [CI = -.506, -

.011]. Since the results did not converge, we conducted further exploratory analyses (two- tailed) at the channel-level across the same time course. We found no evidence of a significant association between HHb concentration difference in Channel 32 and inhibitory performance using Pearson’s (p = .537) or Spearman’s correlations (p = .527).

We did find a significant association between HHb concentration difference in Channel 33 and inhibitory performance using both Pearson’s; r (57) = -.273, p = .036, [CI = -.476, -.056] and Spearman’s tests; rs (57) = -.295, p = .023, [CI = -.519, -.054]. This effect is of a similar, although slightly stronger magnitude to that found at the pair-level, and so suggests that

Channel 33 was likely driving this effect. However, it is important to consider that this weak positive association was identified from exploratory correlational analyses and as such, will require replication in order to confirm its validity. With that caveat in mind, the results of these exploratory analyses suggest that Channel 33 (located in the right anterior/orbital PFC) shows more HHb activation (a larger decrease) in experimental compared to control blocks when infants’ have poorer response inhibition performance (see SM 7e for correlation scatterplots).

32

Discussion

This study aimed to elucidate the neural correlates of response inhibition in 10-month-old infants. We did this by employing fNIRS, a technique that offers a unique insight into the localisation of activation in the awake infant brain, alongside a novel touchscreen task designed to measure response inhibition in infants and toddlers (ECITT; Holmboe et al.,

2020). In this study, we demonstrated that it is possible to successfully measure an index of functional neural activation in infants who are engaged in an active task that requires a motoric response. Furthermore, our results, which align with existing research with pre- schoolers (Mehnert et al., 2013; Moriguchi & Shinohara, 2019; Yanaoka et al., 2020), older children (Cope et al., 2020) and adults (see Aron et al., 2014 for review), provide new evidence for the involvement of right-lateralised regions of the prefrontal and parietal cortices when inhibitory control is exerted already during the first year of life.

The behavioural results of this study confirmed that infants were significantly more accurate when responding to a target in a frequently-rewarded location (prepotent trials) than in an alternative, less frequently-rewarded location (inhibitory trials). This result replicates previous findings indicating that the ECITT is suitable as a measure of inhibitory control in infants (10 months; Hendry et al., 2021), toddlers (16–30 months; Hendry et al., 2021;

Holmboe et al., 2020) and across the lifespan (4 years to adulthood; Holmboe et al., 2020).

From our fNIRS data, we identified a total of six channels covering the right parietal cortex

(Channels 10 and 12), the right DLPFC (Channels 25 and 26), and the right orbital PFC

(Channels 32 and 33) that displayed significantly greater activation (increase in HbO2 and/or decrease in HHb) when the block was inhibition demanding (experimental) compared to when no inhibition was required (control blocks). Our findings are in agreement with a large

(N = 290) longitudinal fMRI study of 7 – 23 year olds which found activation in regions including the orbital frontal cortex, the right DLPFC (Brodmann area 9/46), the inferior

33 frontal gyrus (ventrolateral PFC) and the superior parietal lobule during a Go/No-Go task

(Cope et al., 2020). Although we did not find significant inhibition-specific activation in the ventrolateral PFC in this study, it has been suggested that activation in this area during inhibitory control tasks increases progressively with age (Rubia et al., 2006), and so it is possible that activation in this area would still be weak in 10-month-old infants.

We observed a significantly larger HbO2 concentration increase in experimental compared to control blocks in the pair of channels covering the right DLPFC (Brodmann area 9/46), and that in Channel 25, there was also a significantly larger HHb concentration decrease during experimental compared to control blocks. These results are consistent with previous neuroimaging literature that has found activation in the right DLPFC during inhibitory control tasks across a wide age range, including early to middle childhood (Mehnert et al.,

2013; Moriguchi & Shinohara, 2019; Yanaoka et al., 2020), adolescence (Tamm et al., 2002), and adulthood (Aron et al., 2003, 2004; 2014). Our results also coalesce with early work from

Diamond (1991, 2002; Diamond & Goldman-Rakic, 1989) who demonstrated evidence of a robust link between the DLPFC and performance on the A-not-B task (which measures working memory and inhibitory control) in monkeys, and who predicted a similar association in human infants based on behavioural data.

We observed a significantly larger HbO2 concentration increase in experimental compared to control blocks in the right inferior (Channel 10) and superior (Channel 12) parietal cortex.

Therefore, like older children and adults (Kolodny et al., 2020; Mehnert et al., 2013; van

Belle et al., 2014), 10-month-old infants activate the right parietal cortex when inhibition is required. This finding is consistent with recent evidence that the prefrontal and parietal cortex work in conjunction during early EF development (Fiske & Holmboe, 2019). Finally, activation in channels covering the right orbital PFC was also found, however effects in these channels were not as robust as those observed in the right DLPFC and parietal cortex and

34 were primarily present in the HHb signal. Although there is limited evidence of orbitofrontal involvement in response inhibition in early childhood, previous neuroimaging studies with older children and adults have highlighted the involvement of this region in response inhibition tasks (Casey et al., 1997; Cope et al., 2020; Rubia et al., 2006). The results of the current study suggest that the right orbital PFC plays a role in inhibitory control from as early as 10-months of age.

Against our pre-registered predictions, we did not find a significant association between

HbO2 concentration differences in the right parietal cortex and individual differences in response inhibition performance. This is despite finding that parietal activation (HbO2 increase) was greater in experimental compared to control blocks. These results indicate that individual differences in inhibitory performance were not driven by the magnitude of the

HbO2 response in experimental (relative to control) blocks in parietal regions. In contrast, and as predicted, infants with better response inhibition performance had greater HHb activation

(larger decrease in experimental compared to control blocks) in the DLPFC (Channel 25).

Interestingly, these finding align with fMRI and TMS research by Osada et al. (2019) who also failed to find a significant association between brain activation in the intraparietal sulcus and stop-signal reaction time (a measure of response inhibition) in adult participants, but did observe a significant association with activity in the inferior frontal cortex.

Since the association did not survive correction for multiple comparisons, replication in further studies is needed before we can conclusively say that neural activation (as indexed by the HHb response) in the right lateral PFC is robustly associated with inhibitory performance differences in infancy. Furthermore, the modest size of the effect (r ~ .25) suggests that associations between neural activation and individual response inhibition performance is at best quite weak at this young age. Nevertheless, given the current lack of evidence for associations between inhibitory performance and neural activation in the cortical areas

35 mediating response inhibition in infancy, this preliminary finding will be interesting to follow up in future work.

Results of pre-registered analyses revealed a significant negative association between the

HHb difference in the right orbital PFC and response inhibition performance; further exploratory analyses revealed that Channel 33 was driving this association. As such, infants who showed more HHb activation (in experimental compared to control blocks) in Channel

33 were less able to inhibit their responses. A similar result was obtained by Moriguchi et al.

(2018) who found that pre-schoolers activated the right inferior PFC more when they made an impulsive (compared to a delayed) choice, as such, poorer inhibitory control was associated with more right inferior PFC activation. Moriguchi et al. (2018) hypothesised that pre-schoolers are activating the right inferior PFC because they are recruiting control processes to delay gratification, however fail to do so behaviourally. Although the activation may not be in the exact same location as in Moriguchi et al. (2018), it is possible that infants in the current study are activating the orbital PFC when they are recruiting inhibitory control, but are failing to inhibit their responses behaviourally. However, since this finding only had a small effect size and was the result of exploratory analyses, it will require replication in future work.

As any study, the current study has several strengths and limitations. The multi-channel fNIRS probe used in the study allowed us to measure haemoglobin activity over a substantial area of the brain, sampling the bilateral prefrontal and parietal cortices that have previously been linked to response inhibition. However, it should be acknowledged that there will be areas of the brain, such as the motor cortex, that would also be expected to be active during the ECITT task but were not covered by the probe and so were not discussed here. Multi- channel fNIRS systems (as well as other multi-channel neuroimaging modalities) already face limitations in that there is an increased risk of finding false positive results due to

36 running multiple parallel tests (Singh & Dan, 2006). When using the Benjamini-Hochberg procedure to control the false discovery rate in the analysis of our fNIRS data, the significant block type effects did not survive correction in all but one channel, although this was not unexpected due to the high number of comparisons (n = 42). Therefore, these results will require replication. Further studies that map the neural correlates of response inhibition across the infancy and early childhood period will support the results of the current study.

It is worth considering that the blocked task design used in this study, including the averaging of haemoglobin concentrations across blocks, could dilute some of the individual differences in neural response. In comparison, an event-related design targets the trial-level, capturing the activation associated with each response (inhibitory or non-inhibitory). Although an event- related design is advantageous in many settings, we considered that it was not feasible with

10-month-old infants due to the reliance on long baselines between each response (e.g., 3.8 –

12s in Mehnert et al. (2013)) where infants would be passively viewing the screen. This would not only result in dissipation of the necessary response prepotency built up during the task, it would also present more opportunities for distraction and boredom resulting in additional data attrition. Indeed, a considerable strength of this study is the high number of blocks completed by infants (6 – 14 blocks, equivalent to 36 – 84 individual trials), which also meant that we could apply stringent block inclusion criteria to ensure that data included in analyses had a high signal-to-noise ratio and was a reliable index of neural activation

(Gemignani & Gervain, 2021).

Mapping of the channel locations to the AAL atlas (Shi et al., 2011) using the forward model allowed us to locate brain activation to specific areas of the PFC and parietal cortex (namely, the right inferior and superior parietal gyrus, the right middle frontal gyrus/right DLPFC and the right orbital middle gyrus/orbital PFC) with much more specificity than EEG, which is more commonly used to investigate the neural correlates of early executive function

37 development (Bell & Fox, 1992; Cuevas et al., 2012; Whedon et al., 2020). Whilst the use of subject-specific head models would improve the accuracy of models of light transport

(Cooper et al., 2012; Ferradal et al., 2014; Fu & Richards, 2020), this would require a MRI scan to be obtained from each participant. Given the difficulty and practical limitations of acquiring subject-specific MRI data from infants, this was not feasible in the current study.

Despite this, by using an age-appropriate model of the infant head, a realistic model of light propagation in the infant head was produced which improves the accuracy of the localisation of activation. In addition, the use of an image reconstruction approach enables concentration changes (and, by extension, activation) to be localised on the cortex itself, removing the need to rely on approximate scalp-cortex correspondences to approximate the location of activation (see Frijia et al. (2021) for a recent example in an infant cohort).

In conclusion, this study provides new evidence that, in a large sample of infants, regions in the right lateral PFC and parietal cortex are more active in task conditions that require response inhibition, a finding that is consistent with existing fMRI/fNIRS studies with older children and adults, but has not previously been established in children under 1 year of age.

Associations were found between neural activation (HHb concentration differences) and response inhibition performance in that activation in the right lateral PFC was associated with better response inhibition, whilst activation in orbital regions of the PFC was associated with poorer response inhibition. In using fNIRS alongside a manual response-based task, we have succeeded in lowering the age-boundary at which it is possible to reliably measure the functional neural mechanisms associated with response inhibition as it emerges in the first year of life. This research presents an opportunity for the field to broaden the range of experimental designs that are used alongside fNIRS to probe questions about the neural correlates of infant cognitive, social, motoric, and affective abilities.

38

Data and Code Availability Statement

The data that support the findings of this study and the custom MATLAB scripts used to analyse the fNIRS data is available on the Open Science Framework (OSF) website

[https://osf.io/mv47n/] under a CC-By Attribution 4.0 International license (please cite the current preprint if using any of these materials). The code for the original ECITT task is available on Figshare [https://figshare.com/articles/software/ECITT_Web_App/13258814], and the code for the blocked version of the ECITT used with fNIRS in this study will be uploaded to Figshare upon acceptance. See Holmboe et al.’s (2020) preprint

(https://psyarxiv.com/k7g4a) for details on how to access a demo version of the original behavioural version of the ECITT. The code used to analyse the fNIRS data (HomER2) is available on Github [https://github.com/homer2] and a MATLAB script of the pre-processing stream used in this study is available on OSF [https://osf.io/mv47n/]. The code used to produce the head model and reconstruct images has been developed and released via [http://www.github.com/DOT-HUB and

[https://github.com/liamhywelcj/ReconstructionTenMonthCohortData]. The reconstructed images presented in this paper are also available as a MATLAB figure in the OSF project associated with this work [https://osf.io/mv47n/].

References

Aasted, C. M., Yücel, M. A., Cooper, R. J., Dubb, J., Tsuzuki, D., Becerra, L., Petkov, M. P.,

Borsook, D., Dan, I., & Boas, D. A. (2015). Anatomical guidance for functional near-

infrared spectroscopy: AtlasViewer tutorial. Neurophotonics, 2(2), 020801.

https://doi.org/10.1117/1.nph.2.2.020801

Aron, A. R., Durston, S., Eagle, D. M., Logan, G. D., Stinear, C. M., & Stuphorn, V. (2007).

39

Converging evidence for a fronto-basal-ganglia network for inhibitory control of action

and cognition. Journal of Neuroscience, 27(44), 11860–11864.

https://doi.org/10.1523/JNEUROSCI.3644-07.2007

Aron, A. R., Fletcher, P. C., Bullmore, E. T., Sahakian, B. J., & Robbins, T. W. (2003). Stop-

signal inhibition disrupted by damage to right inferior frontal gyrus in humans. Nature

Neuroscience, 6(2), 115–116. https://doi.org/10.1038/nn1003

Aron, A. R., & Poldrack, R. A. (2006). Cortical and subcortical contributions to stop signal

response inhibition: Role of the subthalamic nucleus. Journal of Neuroscience, 26(9),

2424–2433. https://doi.org/10.1523/JNEUROSCI.4682-05.2006

Aron, A. R., Robbins, T. W., & Poldrack, R. A. (2004). Inhibition and the right inferior

frontal cortex. Trends in cognitive sciences, 8(4), 170-177.

Aron, A. R., Robbins, T. W., & Poldrack, R. A. (2014). Inhibition and the right inferior

frontal cortex: one decade on. Trends in Cognitive Sciences, 18(4), 177–185.

https://doi.org/10.1016/j.tics.2013.12.003

Bell, M. A., & Fox, N. A. (1992). The relations between frontal brain electrical activity and

cognitive development during infancy. Child Development, 63(5), 1142–1163.

https://doi.org/10.1111/j.1467-8624.1992.tb01685.x

Benjamini, Yoav, & Hochberg, Y. (1995). Controlling the false discovery rate: A practical

and powerful approach to multiple testing. Journal of the Royal Statistical Society:

Series B (Methodological), 57(1), 289–300. https://doi.org/10.1111/j.2517-

6161.1995.tb02031.x

Berg, V., Rogers, S. L., McMahon, M., Garrett, M., & Manley, D. (2020). A novel approach

to measure executive functions in students: an evaluation of two child-friendly apps.

40

Frontiers in Psychology, 11(July), 1–15. https://doi.org/10.3389/fpsyg.2020.01702

Bernier, A., Carlson, S. M., & Whipple, N. (2010). From external regulation to self-

regulation: Early parenting precursors of young children’s executive functioning. Child

Development, 81(1), 326–339. https://doi.org/10.1111/j.1467-8624.2009.01397.x

Boas, D. A., & Dale, A. M. (2005). Simulation study of magnetic resonance imaging–guided

cortically constrained diffuse optical tomography of human brain function. Applied

Optics, 44(10), 1957–1968. https://doi.org/10.1364/AO.44.001957

Boas, D. A., Dale, A. M., & Franceschini, M. A. (2004). Diffuse optical imaging of brain

activation: Approaches to optimizing image sensitivity, resolution, and accuracy.

NeuroImage, 23(SUPPL. 1), S275–S288.

https://doi.org/10.1016/j.neuroimage.2004.07.011

Booth, J. R., Burman, D. D., Meyer, J. R., Lei, Z., Trommer, B. L., Davenport, N. D., Li, W.,

Parrish, T. B., Gitelman, D. R., & Mesulam, M. M. (2003). Neural development of

selective attention and response inhibition. NeuroImage, 20(2), 737–751.

https://doi.org/10.1016/S1053-8119(03)00404-X

Brigadoi, S., Galderisi, A., Pieropan, E., Cooper, R. J., Cutini, S., Baraldi, E., Cobelli, C.,

Dell’Acqua, R., Sparacino, G., & Trevisanuto, D. (2019). Mapping hemodynamic

changes during hypoglycemia in the very preterm neonatal brain: Preliminary results.

Optics InfoBase Conference Papers, Part F142-ECBO 2019, 11074_13.

https://doi.org/10.1117/12.2526974

Bulgarelli, C., Blasi, A., de Klerk, C. C. J. M., Richards, J. E., Hamilton, A., & Southgate, V.

(2019). Fronto-temporoparietal connectivity and self-awareness in 18-month-olds: A

resting state fNIRS study. Developmental Cognitive Neuroscience, 38(June), 100676.

https://doi.org/10.1016/j.dcn.2019.100676

41

Bunge, S. A., Dudukovic, N. M., Thomason, M. E., Vaidya, C. J., & Gabrieli, J. D. E. E.

(2002). Immature frontal lobe contributions to cognitive control in children: Evidence

from fMRI. Neuron, 33(2), 301–311. https://doi.org/10.1016/S0896-6273(01)00583-9

Buss, A. T., Fox, N. A., Boas, D. A., & Spencer, J. P. (2014). Probing the early development

of visual working memory capacity with functional near-infrared spectroscopy.

Neuroimage, 85(Pt 1), 314–325. https://doi.org/10.1016/j.neuroimage.2013.05.034

Casey, B. J., Trainor, R. J., Orendi, J. L., Schubert, A. B., Nystrom, L. E., Giedd, J. N.,

Castellanos, F. X., Haxby, J. V, Noll, D. C., Cohen, J. D., Forman, S. D., Dahl, R. E., &

Rapoport, J. L. (1997). A developmental functional mri study of prefrontal activation

during performance of a go-no-go task. Journal of Cognitive Neuroscience, 9(6), 835–

847. https://doi.org/10.1162/jocn.1997.9.6.835

Chikazoe, J., Jimura, K., Hirose, S., Yamashita, K. I., Miyashita, Y., & Konishi, S. (2009).

Preparation to inhibit a response complements response inhibition during performance

of a stop-signal task. Journal of Neuroscience, 29(50), 15870-15877.

Collins-Jones, L. H., Arichi, T., Poppe, T., Billing, A., Xiao, J., Fabrizi, L., Brigadoi, S.,

Hebden, J. C., Elwell, C. E., & Cooper, R. J. (2021). Construction and validation of a

database of head models for functional imaging of the neonatal brain. Human Brain

Mapping, 42(3), 567–586. https://doi.org/10.1002/hbm.25242

Cooper, R. J., Caffini, M., Dubb, J., Fang, Q., Custo, A., Tsuzuki, D., Fischl, B., Wells, W.,

Dan, I., & Boas, D. A. (2012). Validating atlas-guided DOT: A comparison of diffuse

optical tomography informed by atlas and subject-specific anatomies. NeuroImage,

62(3), 1999–2006. https://doi.org/10.1016/j.neuroimage.2012.05.031

Cope, L. M., Hardee, J. E., Martz, M. E., Zucker, R. A., Nichols, T. E., & Heitzeg, M. M.

(2020). Developmental maturation of inhibitory control circuitry in a high-risk sample:

42

A longitudinal fMRI study. Developmental Cognitive Neuroscience, 43, 100781.

https://doi.org/10.1016/j.dcn.2020.100781

Cuevas, K., Swingler, M. M., Bell, M. A., Marcovitch, S., & Calkins, S. D. (2012). Measures

of frontal functioning and the emergence of inhibitory control processes at 10 months of

age. Developmental Cognitive Neuroscience, 2(2), 235–243.

https://doi.org/10.1016/j.dcn.2012.01.002 de Klerk, C. C. J. M., Bulgarelli, C., Hamilton, A., & Southgate, V. (2019). Selective facial

mimicry of native over foreign speakers in preverbal infants. Journal of Experimental

Child Psychology, 183, 33–47. https://doi.org/10.1016/j.jecp.2019.01.015 de Klerk, C. C. J. M., Hamilton, A. F. d. C., & Southgate, V. (2018). Eye contact modulates

facial mimicry in 4-month-old infants: An EMG and fNIRS study. Cortex, 106, 93–103.

https://doi.org/10.1016/j.cortex.2018.05.002

Di Lorenzo, R., Pirazzoli, L., Blasi, A., Bulgarelli, C., Hakuno, Y., Minagawa, Y., &

Brigadoi, S. (2019). Recommendations for motion correction of infant fNIRS data

applicable to multiple data sets and acquisition systems. NeuroImage, 200(June), 511–

527. https://doi.org/10.1016/j.neuroimage.2019.06.056

Diamond, A. (1991). Frontal lobe involvement in cognitive changes during the first year of

life. In K. . Gibson & A. C. Peterson (Eds.), Brain maturation and cognitive

development: Comparative and cross-cultural perspectives (pp. 127–180).

Diamond, A. (2002). Normal development of prefrontal cortex from birth to young

adulthood: Cognitive functions, anatomy, and biochemistry. In D. T. Stuss & R. T.

Knight (Eds.), Principles of frontal lobe function. Oxford University Press.

https://doi.org/10.1093/acprof:oso/9780195134971.003.0029

43

Diamond, A., & Goldman-Rakic, P. S. (1989). Comparison of human infants and rhesus

monkeys on Piaget’s A-not-B task: Evidence for dependence on dorsolateral prefrontal

cortex. Experimental Brain Research, 74, 24–40. https://doi.org/10.1007/BF00248277

Drewe, E. A. (1975). Go - No Go Learning After Frontal Lobe Lesions in Humans. Cortex,

11(1), 8–16. https://doi.org/10.1016/S0010-9452(75)80015-3

Durston, S., Davidson, M. C., Tottenham, N., Galvan, A., Spicer, J., Fossella, J. A., & Casey,

B. J. (2006). A shift from diffuse to focal cortical activity with development.

Developmental Science, 9(1), 1–8. https://doi.org/10.1111/j.1467-7687.2005.00454.x

Durston, S., Thomas, K. M., Yang, Y., Ulu, A. M., Zimmerman, R. D., Casey, B. J., Uluǧ, A.

M., Zimmerman, R. D., & Casey, B. J. (2002). A neural basis for the development of

inhibitory control. Developmental Science, 5(4), F9–F16. https://doi.org/10.1111/1467-

7687.00235

Fang, Q., & Boas, D. A. (2009). Tetrahedral mesh generation from volumetric binary and

grayscale images. Proceedings - 2009 IEEE International Symposium on Biomedical

Imaging: From Nano to Macro, ISBI 2009. https://doi.org/10.1109/ISBI.2009.5193259

Ferradal, S. L., Eggebrecht, A. T., Hassanpour, M., Snyder, A. Z., & Culver, J. P. (2014).

Atlas-based head modeling and spatial normalization for high-density diffuse optical

tomography: In vivo validation against fMRI. NeuroImage, 85, 117–126.

https://doi.org/10.1016/j.neuroimage.2013.03.069

Fiske, A., & Holmboe, K. (2019). Neural substrates of early executive function development.

Developmental Review, 52. https://doi.org/10.1016/j.dr.2019.100866

Friedman, N. P., & Miyake, A. (2004). The relations among inhibition and interference

control functions: a latent-variable analysis. Journal of Experimental Psychology:

44

General, 133(1), 101–135. https://doi.org/10.1037/0096-3445.133.1.101

Friedman, N. P., & Miyake, A. (2017). Unity and diversity of executive functions: Individual

differences as a window on cognitive structure. Cortex, 86, 186–204.

https://doi.org/10.1016/j.cortex.2016.04.023

Friedman, N. P., Miyake, A., Robinson, J. A. L., & Hewitt, J. K. (2011). Developmental

trajectories in toddlers’ self-restraint predict individual differences in executive

functions 14 years later: a behavioral genetic analysis. Developmental Psychology,

47(5), 1410–1430. https://doi.org/10.1037/a0023750

Frijia, E. M., Billing, A., Lloyd-Fox, S., Vidal Rosas, E., Collins-Jones, L., Crespo-Llado, M.

M., Amadó, M. P., Austin, T., Edwards, A., Dunne, L., Smith, G., Nixon-Hill, R.,

Powell, S., Everdell, N. L., & Cooper, R. J. (2021). Functional imaging of the

developing brain with wearable high-density diffuse optical tomography: A new

benchmark for infant neuroimaging outside the scanner environment. NeuroImage, 225.

https://doi.org/10.1016/j.neuroimage.2020.117490

Fu, X., & Richards, J. E. (2020). Age-related changes in diffuse optical tomography

sensitivity to the cortex in infancy. BioRxiv.

https://www.biorxiv.org/content/early/2020/08/23/2020.08.22.262477

Garon, N., Bryson, S. E., & Smith, I. M. (2008). Executive function in preschoolers: A

review using an integrative framework. Psychological Bulletin, 134(1), 31–60.

https://doi.org/10.1037/0033-2909.134.1.31

Garon, N., Smith, I. M., & Bryson, S. E. (2014). A novel executive function battery for

preschoolers: Sensitivity to age differences. Child Neuropsychology, 20(6), 713–736.

https://doi.org/10.1080/09297049.2013.857650

45

Gemignani, J., & Gervain, J. (2021). Comparing different pre-processing routines for infant

fNIRS data. Developmental Cognitive Neuroscience, 48, 100943.

https://doi.org/10.1016/j.dcn.2021.100943

Gonzalez Alam, T., Murphy, C., Smallwood, J., & Jefferies, E. (2018). Meaningful

inhibition: Exploring the role of meaning and modality in response inhibition.

NeuroImage, 181, 108–119. https://doi.org/10.1016/j.neuroimage.2018.06.074

Hampshire, A., Chamberlain, S. R., Monti, M. M., Duncan, J., & Owen, A. M. (2010). The

role of the right inferior frontal gyrus: inhibition and attentional control. NeuroImage,

50(3), 1313–1319. https://doi.org/10.1016/j.neuroimage.2009.12.109

Hendry, A., Greenhalgh, I., Bailey, R., Fiske, A., Dvergsdal, H., & Holmboe, K. (2021).

Development of directed global inhibition, competitive inhibition and behavioural

inhibition during the transition between infancy and toddlerhood. PsyArXiv.

https://doi.org/10.31234/OSF.IO/MHKAJ

Holmboe, K, Csibra, G., & Johnson, M. H. (2008). Frontal cortex functioning in infancy

[University of London]. In Birkbeck: Vol. PhD. https://doi.org/10.31237/osf.io/qe9ck

Holmboe, Karla, Bonneville-Roussy, A., Csibra, G., & Johnson, M. H. (2018). Longitudinal

development of attention and inhibitory control during the first year of life.

Developmental Science, 21(6), e12690. https://doi.org/10.1111/desc.12690

Holmboe, Karla, Larkman, C., de Klerk, C., Simpson, A., Bell, M. A., Patton, L.,

Christodoulou, C., & Dvergsdal, H. (2020). The Early Childhood Inhibitory

Touchscreen Task: A new measure of response inhibition in toddlerhood and across the

lifespan. PsyArXiv, 6 July, https://doi.org/10.31234/osf.io/k7g4a

Howard, S. J., & Melhuish, E. (2017). An early years toolbox for assessing early executive

46

function, language, self-regulation, and social development: Validity, reliability, and

preliminary norms. Journal of Psychoeducational Assessment, 35(3), 255–275.

https://doi.org/10.1177/0734282916633009

Huppert, T. J., Diamond, S. G., Franceschini, M. A., & Boas, D. A. (2009). HomER: A

review of time-series analysis methods for near-infrared spectroscopy of the brain.

Applied Optics, 48(10). https://doi.org/10.1364/AO.48.00D280

Jenkinson, M., Pechaud, M., & Smith, S. (2005). BET2-MR-based estimation of brain, skull

and scalp surfaces. Human Brain Mapping, 17(2), 143–155.

www.fmrib.ox.ac.uk/analysis/research/bet

Kerr-German, A. N., & Buss, A. T. (2020). Exploring the neural basis of selective and

flexible dimensional attention: an fNIRS study. Journal of Cognition and Development,

00(00), 1–13. https://doi.org/10.1080/15248372.2020.1760279

Kim, H.-Y. (2013). Statistical notes for clinical researchers: assessing normal distribution (2)

using skewness and kurtosis. Restorative Dentistry & Endodontics, 38(1), 52.

https://doi.org/10.5395/rde.2013.38.1.52

Kolodny, T., Mevorach, C., Stern, P., Biderman, N., Ankaoua, M., Tsafrir, S., & Shalev, L.

(2020). Fronto-parietal engagement in response inhibition is inversely scaled with

attention-deficit/hyperactivity disorder symptom severity. NeuroImage: Clinical,

25(April 2019), 102119. https://doi.org/10.1016/j.nicl.2019.102119

Lloyd-Fox, S., Blasi, A., & Elwell, C. E. (2010). Illuminating the developing brain: The past,

present and future of functional near infrared spectroscopy. Neuroscience and

Biobehavioral Reviews, 34(3), 269–284.

https://doi.org/10.1016/j.neubiorev.2009.07.008

47

Lloyd-Fox, Sarah, Blasi, A., McCann, S., Rozhko, M., Katus, L., Mason, L., Austin, T.,

Moore, S. E., & Elwell, C. E. (2019). Habituation and novelty detection fNIRS brain

responses in 5- and 8-month-old infants: The Gambia and UK. Developmental Science,

22(5), e12817. https://doi.org/10.1111/desc.12817

Lloyd-Fox, Sarah, Széplaki-Köllod, B., Yin, J., & Csibra, G. (2015). Are you talking to me?

Neural activations in 6-month-old infants in response to being addressed during natural

interactions. Cortex, 70, 35–48. https://doi.org/10.1016/j.cortex.2015.02.005

Mehnert, J, Akhrif, A., Telkemeyer, S., Rossi, S., Schmitz, C. H., Steinbrink, J.,

Wartenburger, I., Obrig, H., & Neufang, S. (2013). Developmental changes in brain

activation and functional connectivity during response inhibition in the early childhood

brain. Brain and Development, 35(10), 894–904.

https://doi.org/10.1016/j.braindev.2012.11.006

Miyake, A., & Friedman, N. P. (2012). The nature and organization of individual differences

in executive functions: Four general conclusions. Current Directions in Psychological

Science, 21(1), 8–14. https://doi.org/10.1177/0963721411429458

Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter, A., & Wager, T. D.

(2000). The unity and diversity of executive functions and their contributions to

complex “frontal lobe” tasks: A latent variable analysis. Cognitive Psychology, 41(1),

49–100. https://doi.org/10.1006/cogp.1999.0734

Monden, Y., Dan, I., Nagashima, M., Dan, H., Uga, M., Ikeda, T., Tsuzuki, D., Kyutoku, Y.,

Gunji, Y., Hirano, D., Taniguchi, T., Shimoizumi, H., Watanabe, E., & Yamagata, T.

(2015). Individual classification of ADHD children by right prefrontal hemodynamic

responses during a go/no-go task as assessed by fNIRS. NeuroImage: Clinical, 9, 1–12.

https://doi.org/10.1016/j.nicl.2015.06.011

48

Moriguchi, Y., & Shinohara, I. (2019a). Less Is More Activation: The involvement of the

lateral prefrontal regions in a “Less Is More” task. Developmental Neuropsychology,

44(3), 273–281. https://doi.org/10.1080/87565641.2019.1608549

Moriguchi, Y., Shinohara, I., & Yanaoka, K. (2018). Neural correlates of delay of

gratification choice in young children: Near-infrared spectroscopy studies.

Developmental Psychobiology, 60(8), 989–998. https://doi.org/10.1002/dev.21791

Mulder, H., Hoofs, H., Verhagen, J., van der Veen, I., & Leseman, P. P. M. M. (2014).

Psychometric properties and convergent and predictive validity of an executive function

test battery for two-year-olds. Frontiers in Psychology, 5(733), 1–17.

https://doi.org/10.3389/fpsyg.2014.00733

Osada, T., Ohta, S., Ogawa, A., Tanaka, M., Suda, A., Kamagata, K., Hori, M., Aoki, S.,

Shimo, Y., Hattori, N., Shimizu, T., Enomoto, H., Hanajima, R., Ugawa, Y., & Konishi,

S. (2019). An essential role of the intraparietal sulcus in response inhibition predicted by

parcellation-based network. Journal of Neuroscience, 39(13), 2509–2521.

https://doi.org/10.1523/JNEUROSCI.2244-18.2019

Petrides, M., & Pandya, D. N. (2012). The Frontal Cortex. In The Human Nervous System

(pp. 988–1011). Elsevier Inc. https://doi.org/10.1016/B978-0-12-374236-0.10026-4

Reyes, L. D., Wijeakumar, S., Magnotta, V. A., Forbes, S. H., & Spencer, J. P. (2020). The

functional brain networks that underlie visual working memory in the first two years of

life. NeuroImage, 116971. https://doi.org/10.1016/j.neuroimage.2020.116971

Rubia, K., Smith, A. B., Woolley, J., Nosarti, C., Heyman, I., Taylor, E., & Brammer, M.

(2006). Progressive increase of frontostriatal brain activation from childhood to

adulthood during event-related tasks of cognitive control. Human Brain Mapping,

27(12), 973–993. https://doi.org/10.1002/hbm.20237

49

Scholkmann, F., & Wolf, M. (2013). General equation for the differential pathlength factor of

the frontal human head depending on wavelength and age. Journal of Biomedical

Optics, 18(10), 105004. https://doi.org/10.1117/1.jbo.18.10.105004

Schweiger, M., & Arridge, S. (2014). The Toast++ software suite for forward and inverse

modeling in optical tomography. Journal of Biomedical Optics, 19(4), 040801.

https://doi.org/10.1117/1.jbo.19.4.040801

Shi, F., Yap, P.-T., Wu, G., Jia, H., Gilmore, J. H., Lin, W., & Shen, D. (2011). Infant brain

atlases from neonates to 1- and 2-year-olds. PLoS ONE, 6(4), e18746.

https://doi.org/10.1371/journal.pone.0018746

Simpson, A., & Riggs, K. J. (2005). Inhibitory and working memory demands of the day-

night task in children. British Journal of Developmental Psychology, 23(3), 471–486.

https://doi.org/10.1348/026151005X28712

Singh, A. K., & Dan, I. (2006). Exploring the false discovery rate in multichannel NIRS.

NeuroImage, 33(2), 542–549. https://doi.org/10.1016/j.neuroimage.2006.06.047

Tamm, L., Menon, V., & Reiss, A. L. (2002). Maturation of brain function associated with

response inhibition. Journal of the American Academy of Child & Adolescent

Psychiatry, 41(10), 1231–1238. https://doi.org/10.1097/00004583-200210000-00013 van Belle, J., Vink, M., Durston, S., & Zandbelt, B. B. (2014). Common and unique neural

networks for proactive and reactive response inhibition revealed by independent

component analysis of functional MRI data. NeuroImage, 103, 65–74.

https://doi.org/10.1016/j.neuroimage.2014.09.014

Wager, T. D., Sylvester, C. Y. C., Lacey, S. C., Nee, D. E., Franklin, M., & Jonides, J.

(2005). Common and unique components of response inhibition revealed by fMRI.

50

NeuroImage, 27(2), 323–340. https://doi.org/10.1016/j.neuroimage.2005.01.054

Whedon, M., Perry, N. B., & Bell, M. A. (2020). Relations between frontal EEG maturation

and inhibitory control in preschool in the prediction of children’s early academic skills.

Brain and Cognition, 146(October), 105636.

https://doi.org/10.1016/j.bandc.2020.105636

Wiebe, S. A., Sheffield, T. D., & Espy, K. A. (2012). Separating the fish from the sharks: A

longitudinal study of preschool response inhibition. Child Development, 83(4), 1245–

1261. https://doi.org/10.1111/j.1467-8624.2012.01765.x

Wijeakumar, S., Kumar, A., M. Delgado Reyes, L., Tiwari, M., & Spencer, J. P. (2019).

Early adversity in rural India impacts the brain networks underlying visual working

memory. Developmental Science, 22(5), e12822. https://doi.org/10.1111/desc.12822

Wilcox, T., & Biondi, M. (2015). fNIRS in the developmental sciences. Wiley

Interdisciplinary Reviews: Cognitive Science, 6(3), 263–283.

https://doi.org/10.1002/wcs.1343

Yanaoka, K., Moriguchi, Y., & Saito, S. (2020a). Cognitive and neural underpinnings of goal

maintenance in young children. Cognition, 203(October 2019), 104378.

https://doi.org/10.1016/j.cognition.2020.104378

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. Monographs of the Society for Research in Child Development, 78(4), 16–33.

51