The adult brain is highly specialised, with core networks coordinating to subserve complex behaviours. This specialised functioning emerges across development through a combination of genetically influenced brain architecture and experience-expectant learning processes (generalised neural development that occurs as a result of common experiences) and experience-dependent (variation in the environment contributing to individual differences in neural response) (Johnson, 2001). During early development, infants undergo significant neural, physiological, and socio-cognitive changes that are accompanied by large-scale changes in social communication and interaction. Currently, we have relatively few tools that allow us to comprehensively capture the emergence of functional specialisation in the infant social brain. Developing new approaches is critical for advancing our understanding of early brain physiology and cognitive function.

Identifying appropriate metrics to index functional specialisation in the infant brain should be informed by theoretical perspectives on how functional specialisation develops. Interactive specialisation is a theory of brain development that posits that functional specialisation emerges through competition between brain regions (Johnson, 2011). Thus, functional specialisation can be indexed as a smaller spatial extent of neural responses to a particular stimulus category and concomitant selectivity in responsive regions (Jones et al., 2015). Typically, the extent and selectivity of brain activation is measured through indirect indices of oxygenated blood flow (e.g. functional near-infrared spectroscopy [fNIRS] [Bale et al., 2016] or functional magnetic resonance imaging [fMRI] [Attwell and Iadecola, 2002]) or of coordinated neural activity (e.g. electroencephalography [EEG] [Buzsáki, 2006]). However, one mechanism that may contribute to competition between brain regions is the limited energetic resources available to the infant brain. The brain is an energetically costly organ, consuming 20–25% of the body’s energy in adulthood while representing only 2% of the body’s mass (Raichle and Mintun, 2006; Sokoloff, 1999). There are also substantial developmental changes in the brain’s energy consumption; in the first year of life, up to 60% of available energy is used by the brain (Steiner, 2019). When brain regions become functionally active (e.g. during stimulus processing) neurons fire more rapidly, requiring greater supplies of adenosine triphosphate (ATP) (energy stores). Producing ATP requires oxygen, and this is supplied through a localised increase in oxygenated haemoglobin in the blood. Increases in oxygenated haemoglobin do not happen concurrently in all brain areas, and there are spatial dependencies between activated and deactivated regions in the adult brain (Leech et al., 2014). Energy supplies are important to synaptic plasticity, memory, and learning (Vaynman et al., 2006), and the mechanism through which energy supplies are coupled to activation (neurovascular coupling, see ) also develops through experience-dependent specialisation in the infant brain (Kozberg and Hillman, 2016). Thus, energy supply constraints may be one factor that contributes to the emergence of brain specialisation. If this is the case, detecting functional specialisation in infancy requires not only examining measures of neural activity and oxygenated haemoglobin, but also identifying whether particular regions show stronger coupling between neuronal demand and energetic supply.

As a first step, testing such frameworks requires the availability of methods that can measure the spatial extent and stimulus selectivity of neuroenergetics coupling in infancy. Previous studies have typically used single modalities sensitive to distinct aspects of brain function. For example, studies with fMRI indicate that core regions of the social brain (particular the fusiform face area) show increases in oxygenated haemoglobin delivery in response to faces by 4–9 months (Kosakowski et al., 2022). Further, fNIRS studies show that oxygenated haemoglobin delivery in response to naturalistic social videos in a broad region of temporal cortex emerges over the first hours of life (Farroni et al., 2013). Work with EEG indicates developmental increases in differentiated theta power responses to social versus non-social stimuli between 6 and 12 months (Jones et al., 2015). Thus, work with single modalities indicates development in functional specialisation across the first year of life.

Broadband near-infrared spectroscopy (bNIRS) is a new technique that uses a broad range of optical wavelengths which allows the measurement of the oxidation state of mitochondrial respiratory chain enzyme cytochrome c oxidase (CCO), thereby providing a direct measure of cellular energy metabolism (Bale et al., 2016). CCO is located in the inner mitochondrial membrane and serves as the terminal electron acceptor in the electron transport chain (ETC). It therefore accounts for 95% of cellular oxygen metabolism. In this way, bNIRS allows non-invasive measurement of cellular energy metabolism alongside haemodynamics/oxygenation in awake infants.

Work with single modalities has demonstrated that social selectivity in core regions of the adult ‘social brain’ can be robustly detected by 4–7 months of age (Grossmann et al., 2010; Lloyd-Fox et al., 2012; Lloyd-Fox et al., 2013; Lloyd‐Fox et al., 2018). We recently showed the feasibility of using bNIRS in 4- to 7-month-old typically developing infants (Siddiqui et al., 2021) and demonstrated the presence of unique task-relevant, regionally specific functional networks where high levels of haemodynamic and metabolic coupling were observed. Here, we integrate this methodology with EEG to examine whether specific brain regions show coordinated energetic coupling and neural activity. We develop a novel analysis pipeline to identify localised coupling responses that are modulated by naturalistic social content. We aimed specifically to investigate the relationship between low- and high-frequency neural activity with haemodynamics and metabolism. For EEG, we expected an increase in neural activity in response to the social condition and a decrease in neural activity in response to the non-social condition. Based on previous work, this was expected to be strongest in the theta frequency band (Jones et al., 2015). Moreover, for the combined bNIRS-EEG analyses, we hypothesised differentiated haemodynamic/metabolic coupling with neural activity for the social and non-social stimulus conditions. We performed two types of statistical tests: (a) individual comparisons of the social and non-social conditions and (b) comparison of the social condition versus the non-social condition. The individual condition tests were performed to show the scale and spatial location/sensitivity of the coupling between haemodynamics/metabolism and neural activity for each condition. Meanwhile, the social versus non-social comparison was performed to show where there was a significant difference in the coupling between the two conditions. With comparison (a) we aimed to identify regions involved in the processing of social and non-social stimuli by identifying the regions where the coupling was significant. With comparison (b) we aimed to identify regions where coupling was significantly different between conditions. We predicted that for the individual comparison of the social condition, we would observe positive associations between bNIRS and EEG measures, that is a simultaneous increase in haemodynamics/metabolism and neural oscillatory activity in the beta and gamma frequency bands (based on previous combined EEG-fMRI studies Scheeringa et al., 2011; Scheeringa et al., 2009; Goldman et al., 2002; Yuan et al., 2010; Niessing et al., 2005; Logothetis et al., 2001; Koch et al., 2009) which would be localised to core social brain regions. We hypothesised that for the non-social condition, over the same brain regions, positive associations would be observed between bNIRS and EEG measures, but they would be a simultaneous decrease in haemodynamics/metabolism and oscillatory activity. We also expected simultaneous increases in haemodynamics/metabolism and oscillatory activity to be localised to the parietal brain region. These predictions are based on our previous work (Siddiqui et al., 2021) where we demonstrated that stronger coupling between haemodynamics and metabolism was observed in the temporo-parietal regions for the social condition and in parietal region for the non-social condition which is known to play an important role in object processing (Wilcox et al., 2010; Dekker et al., 2011). For the social versus the non-social contrast, we predicted that haemodynamic activity and metabolism would be coupled with neuronal oscillatory activity more strongly for the social stimuli in comparison to the non-social stimuli, with significant differences being observed in the temporo-parietal regions.

We develop a tool that enabled multimodal imaging analysis of coordinated neural activation, metabolic demand, and oxygenated haemoglobin delivery in the infant brain. As a proof of principle, we examined the relationship between these measures to identify regional selectivity to social versus non-social stimuli. To first demonstrate the scale and spatial sensitivity of the coupling between haemodynamic/metabolic activity and neuronal oscillatory activity, comparisons were performed individually for the social and non-social conditions. For this, we predicted a simultaneous increase in haemodynamics/metabolism and neural activity in the beta and gamma frequency band. We predicted that for the social condition this would be localised to the core social brain regions (temporo-parietal region) while for the non-social condition, we expected the coupling to be localised to parietal regions, known to be involved in object processing (Wilcox et al., 2010; Dekker et al., 2011). We additionally expected a simultaneous decrease in haemodynamic/metabolic activity and neural activity over the temporo-parietal region for the non-social condition, in accordance with our previous work (Siddiqui et al., 2021). Next, to demonstrate differential coupling for social and non-social stimuli, we performed a comparison of the social condition versus the non-social condition. For this, we hypothesised that in the beta and gamma frequency bands, there would be stronger coupling between haemodynamics/metabolism and neural activity for the social condition over the temporo-parietal region.

Confirming previous work, naturalistic social and non-social stimuli produce broad haemodynamic changes, with smaller spatial extent of locations with coupled haemodynamic and metabolic activity (Siddiqui et al., 2021). We also replicated previously observed greater EEG responses to social versus non-social stimuli (Jones et al., 2015). However, examining coupling between these two phenomena uncovered a precise pattern in which specific locations in the parietal and temporo-parietal regions showed differential coupling between bNIRS-EEG for social and non-social stimuli, particularly for the beta and gamma band frequency bands, as we predicted. We contend that this approach identifies a more localised regional area with selective coordination of neural, haemodynamic, and metabolic activity. The increased localisation observed in our coupling analysis may indicate our approach provides a more rigorous measure of functional specialisation. Widespread use of this technique will accelerate our understanding of both the typically and atypically developing brain. Unexpectedly, while most associations between haemodynamic/metabolic activity and oscillatory activity were localised, we observed several long-range connections between haemodynamic/metabolic and neural signals. It has been hypothesised that long-range functional connectivity patterns are vital for the organisation of human brain structure and function (Wang et al., 2021). The strongest coupling was observed between temporo-parietal bNIRS channel 14 with parietal EEG locations Pz and PO4 for the social condition (for beta and gamma frequency bands). Meanwhile, for the non-social condition, coupling was observed between temporo-parietal bNIRS channel 14 with occipital and parietal EEG locations Oz, O2, PO8, and P10 (for theta and beta frequency bands). While an overall consistent pattern of associations across chromophores and conditions was observed, some variability was also seen, particularly across frequency bands. This was expected and in line with previous EEG-fMRI studies that have demonstrated task-dependent variation in coupling between neural and haemodynamic activity across frequency bands (Scheeringa et al., 2011; Scheeringa et al., 2009; Goldman et al., 2002; Yuan et al., 2010; Niessing et al., 2005; Logothetis et al., 2001; Koch et al., 2009). For example, for resting-state simultaneous fMRI and EEG, stronger coupling between the BOLD response and neural activity has been observed for the alpha band (Scheeringa et al., 2012). Meanwhile, for cognitive tasks, stronger coupling has been observed in the gamma frequency band (Kucewicz et al., 2014). Scheeringa et al., 2011, investigated trial-by-trial coupling of EEG and BOLD activity and found that low- and high-frequency bands independently contribute to explaining BOLD variance. We therefore expected the frequency band showing the strongest coupling between bNIRS and EEG for each of the stimuli to vary. Further, while we did expect and observe significant overlap in associations between chromophores within each frequency band, some variability was seen. For example, for the social condition, no associations were observed in the low-frequency bands for any of the chromophores. Moreover, in the beta frequency bands, all chromophores displayed significant associations between bNIRS channel 14 and Pz for the social condition and both HHb and oxCCO displayed significant associations between bNIRS channel 14 and O2, PO8, and C2. Similarly, in the gamma frequency bands, both HbO₂ and oxCCO displayed significant associations between bNIRS channel 14 and PO4. The variability that was observed between chromophores was limited mostly to the non-social condition. For example, only oxCCO displayed significant associations between bNIRS and EEG for the low-frequency theta and alpha frequency bands. It is well known that various components involved in neurovascular coupling undergo development postnatally, see the review by Harris et al., 2011, for a full discussion. Briefly, there is an extensive structural change within cerebral microvasculature including growth, extension, and proliferation of new blood vessels (Rowan and Maxwell, 1981; Norman and O’Kusky, 1986). Further, studies have also demonstrated gradual development of vascular reactivity (i.e. change in vascular tone, vasoconstriction, and vasodilation) (Scheeringa et al., 2011; Scheeringa et al., 2009) which is necessary for the propagation of the NVC response (Chen et al., 2014). Lastly, pericytes and astrocytes which are key components of NVC are also known to undergo development in size, number, connectivity, and branching (Kozberg and Hillman, 2016; Fujimoto, 1995; Seregi et al., 1987; Stichel et al., 1991). From the metabolic perspective, infant positron emission tomography (PET) studies demonstrate regional-specific, progressive increase in the cerebral metabolic rate of oxygen consumption (CMRO₂) (Chugani et al., 1987) while others evidence a developmental maturational change in oxidative metabolism (Kozberg and Hillman, 2016). In adults, previous research has also suggested that oxygen consumption is more spatially localised in comparison to changes in cerebral blood flow (Malonek and Grinvald, 1996) and that oxCCO has distinct spatial distributions in the brain (Phan et al., 2016; Wong-Riley et al., 1993; Hevner et al., 1995), indicating that energy metabolism may be more spatially specific. The spatial distribution of oxCCO in different brain regions currently remains unmapped in the developing infant brain, however. Therefore, taken together, given that during early development there are extensive changes in cerebral vasculature as well as the metabolic environment and potential variability in the spatial distribution of oxCCO, it is expected that there will be some variability observed in the associations between the haemodynamics and metabolism with neural activity. In our study, we observed more consistent oxCCO-EEG associations across frequency bands and stimuli with more localised (fewer long-range) associations. Further studies with a larger sample size and longitudinal follow-up can provide a clearer view on how NVC develops in infancy which will help explain some of the observed variability. Moreover, future studies with high-density bNIRS arrays will provide clarification on the spatial distribution of oxCCO in the infant brain.

EEG profiles observed in the present study are consistent with previous studies in identifying increased gamma band activity over temporal and parieto-occipital brain regions during face processing (Uono et al., 2017; Ghuman et al., 2014; Bayer et al., 2018; Müller-Bardorff et al., 2018; Nguyen and Cunnington, 2014; Nguyen et al., 2014; Engell and McCarthy, 2011; Bossi et al., 2020; Anaki et al., 2007; Ishai et al., 2000; Pelphrey et al., 2005; Sato et al., 2014; Zion-Golumbic et al., 2008; Gao et al., 2013). High-frequency neural firing is associated with localised processing (von Stein and Sarnthein, 2000) whilst lower-frequency activity is associated with larger-scale network changes and transfer of information across systems (Canolty and Knight, 2010). The increase in lower-frequency activity during social attention also observed here and in other work (Jones et al., 2015; van der Velde et al., 2021) may support larger-scale connectivity and communication of information through cross-frequency coupling (Ghuman et al., 2014). Our work further indicates that measures of metabolic load may provide important additional information in understanding localisation of brain function. Localised high-frequency activity exerts strong metabolic demand (Smith et al., 2002; Kann, 2011) and subsequent increases in oxygenated haemoglobin (Logothetis et al., 2001; Niessing et al., 2005; Goense and Logothetis, 2008). These increases in metabolic rate are supported by increased activity in the mitochondrial ETC, resulting in the changes in CCO we detected with broadband NIRS. Nitric oxide (which competes with oxygen to bind to CCO) and carbon dioxide (produced as a by-product in the ETC) are key signalling molecule in controlling neurovascular coupling and thus subsequent oxygen delivery (Hosford and Gourine, 2019; Hosford et al., 2022). Finally, reactive oxygen species produced by the ETC are a key signal in inducing synaptic plasticity (Oswald et al., 2018). Thus, our work is consistent with a model in which social attention induces localised high-frequency brain activity in the temporal parietal junction, which increases local metabolic rates, triggering synaptic plasticity and subsequent oxygen delivery to a broader region.

Our work particularly highlights the temporal-parietal junction is showing strong coupling and social selectivity. Previous studies measuring haemodynamic activity have identified early sensitivity of this region to social stimuli from at least 4 months (Lloyd-Fox et al., 2017), alongside a broader network of other regions. Here, we pinpoint this specific location as having coupled neuronal, metabolic, and haemodynamic activity that is modulated in opposite directions by complex social and non-social content. In the adult brain, the temporal-parietal junction has received considerable attention and there are several competing models of its function. It has been linked to mentalising (Schurz et al., 2014; Schurz et al., 2017) and reorienting attention to behaviourally relevant stimuli (Corbetta and Shulman, 2002); it can be viewed as a nexus area where the convergence of attention, language, memory, and social processing supports a social context for behaviour (Carter and Huettel, 2013) or as a region that is active when awareness of a prediction permits attentional control (Wilterson et al., 2021). Intriguingly, recent formulations within the predictive coding framework link the right temporal-parietal junction to a domain-general role in prediction, perhaps representing the precision of priors (Masina et al., 2022). Predictability has been linked to energy efficiency, with some computational models showing that energy limitations are the only requirement for driving the emergence of predictive coding (Ali et al., 2021). Increases in beta/gamma have also been linked to unexpected reward processing (HajiHosseini et al., 2012). Taken together, our results may indicate the early presence of priors for social interaction that are being actively updated (in contrast to the dynamic toys, which may already be more predictable).

The methods we developed could be broadly applied to study both neurotypical and atypical brain function. Assessing coupling over developmental time may reveal the mechanisms underpinning neural specialisation and constrain theoretical frameworks seeking to explain specialisation in the adult brain. The mechanisms of neurovascular coupling remain unclear in the adult brain (Hosford and Gourine, 2019), and are developing in infancy (Kozberg and Hillman, 2016), and novel multimodal and non-invasive approaches to their identification could yield significant progress. Computational models could test the role of constraints in energy supply on developing localisation of function. Further, the region identified here also shows atypical haemodynamic responsiveness in infants with later symptoms of autism (Lloyd‐Fox et al., 2018); since mitochondrial dysfunction has become an increasing focus in autism (Siddiqui et al., 2016) the possibility that atypical coupling may impact specialisation in autism is an important hypothesis to test. Further, our methods have applicability in determining the impacts of early brain injury. Recent work (Bale et al., 2019) measured both cerebral oxygenation and energy metabolism in neonates with brain injury (hypoxic-ischaemic encephalopathy) and demonstrated that the relationship between metabolism and oxygenation was able to predict injury severity. This therefore provided a clinical, non-invasive biomarker of neonatal brain injury. Indicating applicability across the lifespan (Vezyroglou et al., 2022) simultaneous measurements of cerebral oxygenation, metabolism, and neural activity in epilepsy revealed unique metabolic profiles for healthy brain regions in comparison to those with the regions of the epileptic focus. The work in epilepsy demonstrates the strength of combining measurements from multiple modalities to investigate brain states, particularly in clinical populations.

Our work has several limitations. We used naturalistic stimuli to maximise ecological validity; however, this reduces our ability to probe the function of the temporal-parietal junction across specific stimulus dimensions and this is an important target for future work. Limitations of current technology meant we recorded from the right hemisphere only and thus cannot determine the specificity of our findings to left temporal-parietal junction; engineering advances are required to produce whole-head bNIRS devices. Moreover, we only studied one age group of infants between 4 and 7 months; therefore, we could not investigate developmental change.

The data contains human subject data from minors and guardians provided informed consent to having data shared only with researchers involved in the project, in anonymised form. A Patient and Public Involvement (PPI) initiative at the Centre for Brain and Cognitive Development aimed to actively work in partnership with parents and guardians participating in research studies to help design and manage future research. A comprehensive public survey was conducted as part of this initiative which aimed to evaluate parent attitudes to data sharing in developmental science. This survey revealed that majority of parents do not want their data to be shared openly but are open to the data being shared with other researchers related to the project. Therefore, in order to adhere to participant preference/choice, a curated data sharing approach must be followed wherein the data can only be made available upon reasonable request through a formal data sharing and project affiliation agreement. The researcher will have to contact MFS via email on m.siddiqui@bbk.ac.uk and complete a project affiliation form providing their study aims, a detailed study proposal, plan for the analysis protocol, ethics, and plans for data storage and protection. Successful proposals will have aims aligned with the aims of the original study. Raw NIRS data, EEG data and integrated NIRS-EEG data can be made available in anonymised form. ID numbers linking the NIRS and EEG data, however, cannot be provided as parents/guardians have consented only to data being shared in anonymised form. All code used to analyse the NIRS data and the integration of the NIRS and EEG data is available on GitHub (https://github.com/maheensiddiqui91/NIRS-EEG, copy archived at Siddiqui, 2023). EEG data was processed using EEGlab which is a publicly available toolbox.

1. 1. Ali A
   2. Ahmad N
   3. de Groot E
   4. van Gerven MAJ
   5. Kietzmann TC

   (2021) Predictive coding is a consequence of energy efficiency in recurrent neural networks

   SSRN Electronic Journal 1:17.

   https://doi.org/10.2139/ssrn.3976481

   • Google Scholar
2. 1. Anaki D
   2. Zion-Golumbic E
   3. Bentin S

   (2007) Electrophysiological neural mechanisms for detection, configural analysis and recognition of faces

   NeuroImage 37:1407–1416.

   https://doi.org/10.1016/j.neuroimage.2007.05.054

   • PubMed
   • Google Scholar
3. 1. Arifler D
   2. Zhu T
   3. Madaan S
   4. Tachtsidis I

   (2015) Optimal wavelength combinations for near-infrared spectroscopic monitoring of changes in brain tissue hemoglobin and cytochrome c oxidase concentrations

   Biomedical Optics Express 6:933–947.

   https://doi.org/10.1364/BOE.6.000933

   • PubMed
   • Google Scholar
4. 1. Attwell D
   2. Iadecola C

   (2002) The neural basis of functional brain imaging signals

   Trends in Neurosciences 25:621–625.

   https://doi.org/10.1016/s0166-2236(02)02264-6

   • PubMed
   • Google Scholar
5. 1. Bale G
   2. Elwell CE
   3. Tachtsidis I

   (2016) Errata: From Jöbsis to the present day: a review of clinical near-infrared spectroscopy measurements of cerebral cytochrome-c-oxidase

   Journal of Biomedical Optics 21:099801.

   https://doi.org/10.1117/1.JBO.21.9.099801

   • PubMed
   • Google Scholar
6. 1. Bale G
   2. Mitra S
   3. de Roever I
   4. Sokolska M
   5. Price D
   6. Bainbridge A
   7. Gunny R
   8. Uria-Avellanal C
   9. Kendall GS
   10. Meek J
   11. Robertson NJ
   12. Tachtsidis I

   (2019) Oxygen dependency of mitochondrial metabolism indicates outcome of newborn brain injury

   Journal of Cerebral Blood Flow and Metabolism 39:2035–2047.

   https://doi.org/10.1177/0271678X18777928

   • PubMed
   • Google Scholar
7. 1. Bayer M
   2. Rubens MT
   3. Johnstone T

   (2018) Simultaneous EEG-fMRI reveals attention-dependent coupling of early face processing with a distributed cortical network

   Biological Psychology 132:133–142.

   https://doi.org/10.1016/j.biopsycho.2017.12.002

   • PubMed
   • Google Scholar
8. 1. Benjamini Y
   2. Hochberg Y

   (1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing

   Journal of the Royal Statistical Society 57:289–300.

   https://doi.org/10.1111/j.2517-6161.1995.tb02031.x

   • Google Scholar
9. 1. Bevilacqua F
   2. Piguet D
   3. Marquet P
   4. Gross JD
   5. Tromberg BJ
   6. Depeursinge C

   (1999) In vivo local determination of tissue optical properties: applications to human brain

   Applied Optics 38:4939–4950.

   https://doi.org/10.1364/ao.38.004939

   • PubMed
   • Google Scholar
10. 1. Bortfeld H
    2. Wruck E
    3. Boas DA

    (2007) Assessing infants’ cortical response to speech using near-infrared spectroscopy

    NeuroImage 34:407–415.

    https://doi.org/10.1016/j.neuroimage.2006.08.010

    • PubMed
    • Google Scholar
11. 1. Bossi F
    2. Premoli I
    3. Pizzamiglio S
    4. Balaban S
    5. Ricciardelli P
    6. Rivolta D

    (2020) Theta- and gamma-band activity discriminates face, body and object perception

    Frontiers in Human Neuroscience 14:74.

    https://doi.org/10.3389/fnhum.2020.00074

    • PubMed
    • Google Scholar
12. 1. Brigadoi S
    2. Phan P
    3. Highton D
    4. Powell S
    5. Cooper RJ
    6. Hebden J
    7. Smith M
    8. Tachtsidis I
    9. Elwell CE
    10. Gibson AP

    (2017) Image reconstruction of oxidized cerebral cytochrome C oxidase changes from broadband near-infrared spectroscopy data

    Neurophotonics 4:021105.

    https://doi.org/10.1117/1.NPh.4.2.021105

    • PubMed
    • Google Scholar
13. Book

    1. Buzsáki G

    (2006) Rhythms of the Brain

    Oxford University Press.

    https://doi.org/10.1093/acprof:oso/9780195301069.001.0001

    • Google Scholar
14. 1. Canolty RT
    2. Knight RT

    (2010) The functional role of cross-frequency coupling

    Trends in Cognitive Sciences 14:506–515.

    https://doi.org/10.1016/j.tics.2010.09.001

    • PubMed
    • Google Scholar
15. 1. Carter RM
    2. Huettel SA

    (2013) A nexus model of the temporal-parietal junction

    Trends in Cognitive Sciences 17:328–336.

    https://doi.org/10.1016/j.tics.2013.05.007

    • PubMed
    • Google Scholar
16. 1. Chen BR
    2. Kozberg MG
    3. Bouchard MB
    4. Shaik MA
    5. Hillman EMC

    (2014) A critical role for the vascular endothelium in functional neurovascular coupling in the brain

    Journal of the American Heart Association 3:e000787.

    https://doi.org/10.1161/JAHA.114.000787

    • PubMed
    • Google Scholar
17. 1. Chugani HT
    2. Phelps ME
    3. Mazziotta JC

    (1987) Positron emission tomography study of human brain functional development

    Annals of Neurology 22:487–497.

    https://doi.org/10.1002/ana.410220408

    • PubMed
    • Google Scholar
18. 1. Corbetta M
    2. Shulman GL

    (2002) Control of goal-directed and stimulus-driven attention in the brain

    Nature Reviews. Neuroscience 3:201–215.

    https://doi.org/10.1038/nrn755

    • PubMed
    • Google Scholar
19. 1. Corlu A
    2. Choe R
    3. Durduran T
    4. Lee K
    5. Schweiger M
    6. Arridge SR
    7. Hillman EMC
    8. Yodh AG

    (2005) Diffuse optical tomography with spectral constraints and wavelength optimization

    Applied Optics 44:2082–2093.

    https://doi.org/10.1364/ao.44.002082

    • PubMed
    • Google Scholar
20. 1. Custo A
    2. Wells WM
    3. Barnett AH
    4. Hillman EMC
    5. Boas DA

    (2006) Effective scattering coefficient of the cerebral spinal fluid in adult head models for diffuse optical imaging

    Applied Optics 45:4747–4755.

    https://doi.org/10.1364/ao.45.004747

    • PubMed
    • Google Scholar
21. 1. Dekker T
    2. Mareschal D
    3. Sereno MI
    4. Johnson MH

    (2011) Dorsal and ventral stream activation and object recognition performance in school-age children

    NeuroImage 57:659–670.

    https://doi.org/10.1016/j.neuroimage.2010.11.005

    • PubMed
    • Google Scholar
22. 1. Duncan A
    2. Meek JH
    3. Clemence M
    4. Elwell CE
    5. Tyszczuk L
    6. Cope M
    7. Delpy DT

    (1995) Optical pathlength measurements on adult head, calf and forearm and the head of the newborn infant using phase resolved optical spectroscopy

    Physics in Medicine and Biology 40:295–304.

    https://doi.org/10.1088/0031-9155/40/2/007

    • PubMed
    • Google Scholar
23. 1. Engell AD
    2. McCarthy G

    (2011) The relationship of γ oscillations and face-specific ERPs recorded subdurally from occipitotemporal cortex

    Cerebral Cortex 21:1213–1221.

    https://doi.org/10.1093/cercor/bhq206

    • PubMed
    • Google Scholar
24. Conference

    1. Fang Q
    2. Boas D

    (2009) Tetrahedral mesh generation from volumetric binary and gray-scale images

    IEEE International Symposium on Biomedical Imaging. pp. 1142–1145.

    https://doi.org/10.1109/ISBI.2009.5193259

    • Google Scholar
25. 1. Farroni T
    2. Chiarelli AM
    3. Lloyd-Fox S
    4. Massaccesi S
    5. Merla A
    6. Di Gangi V
    7. Mattarello T
    8. Faraguna D
    9. Johnson MH

    (2013) Infant cortex responds to other humans from shortly after birth

    Scientific Reports 3:2851.

    https://doi.org/10.1038/srep02851

    • PubMed
    • Google Scholar
26. 1. Friston KJ
    2. Holmes AP
    3. Worsley KJ
    4. Poline JP
    5. Frith CD
    6. Frackowiak RSJ

    (1994) Statistical parametric maps in functional imaging: A general linear approach

    Human Brain Mapping 2:189–210.

    https://doi.org/10.1002/hbm.460020402

    • Google Scholar
27. 1. Fujimoto K

    (1995) Pericyte-endothelial gap junctions in developing rat cerebral capillaries: A fine structural study

    The Anatomical Record 242:562–565.

    https://doi.org/10.1002/ar.1092420412

    • PubMed
    • Google Scholar
28. 1. Gao Z
    2. Goldstein A
    3. Harpaz Y
    4. Hansel M
    5. Zion-Golumbic E
    6. Bentin S

    (2013) A magnetoencephalographic study of face processing: M170, gamma-band oscillations and source localization

    Human Brain Mapping 34:1783–1795.

    https://doi.org/10.1002/hbm.22028

    • PubMed
    • Google Scholar
29. 1. Ghuman AS
    2. Brunet NM
    3. Li Y
    4. Konecky RO
    5. Pyles JA
    6. Walls SA
    7. Destefino V
    8. Wang W
    9. Richardson RM

    (2014) Dynamic encoding of face information in the human fusiform gyrus

    Nature Communications 5:5672.

    https://doi.org/10.1038/ncomms6672

    • PubMed
    • Google Scholar
30. 1. Goense JBM
    2. Logothetis NK

    (2008) Neurophysiology of the BOLD fMRI signal in awake monkeys

    Current Biology 18:631–640.

    https://doi.org/10.1016/j.cub.2008.03.054

    • PubMed
    • Google Scholar
31. 1. Goldman RI
    2. Stern JM
    3. Engel J
    4. Cohen MS

    (2002) Simultaneous EEG and fMRI of the alpha rhythm

    Neuroreport 13:2487–2492.

    https://doi.org/10.1097/01.wnr.0000047685.08940.d0

    • PubMed
    • Google Scholar
32. 1. Grossmann T
    2. Oberecker R
    3. Koch SP
    4. Friederici AD

    (2010) The developmental origins of voice processing in the human brain

    Neuron 65:852–858.

    https://doi.org/10.1016/j.neuron.2010.03.001

    • PubMed
    • Google Scholar
33. 1. HajiHosseini A
    2. Rodríguez-Fornells A
    3. Marco-Pallarés J

    (2012) The role of beta-gamma oscillations in unexpected rewards processing

    NeuroImage 60:1678–1685.

    https://doi.org/10.1016/j.neuroimage.2012.01.125

    • PubMed
    • Google Scholar
34. 1. Harris JJ
    2. Reynell C
    3. Attwell D

    (2011) The physiology of developmental changes in BOLD functional imaging signals

    Developmental Cognitive Neuroscience 1:199–216.

    https://doi.org/10.1016/j.dcn.2011.04.001

    • PubMed
    • Google Scholar
35. 1. Hevner RF
    2. Liu S
    3. Wong-Riley MTT

    (1995) A metabolic map of cytochrome oxidase in the rat brain: histochemical, densitometric and biochemical studies

    Neuroscience 65:313–342.

    https://doi.org/10.1016/0306-4522(94)00514-6

    • PubMed
    • Google Scholar
36. 1. Hosford PS
    2. Gourine AV

    (2019) What is the key mediator of the neurovascular coupling response?

    Neuroscience and Biobehavioral Reviews 96:174–181.

    https://doi.org/10.1016/j.neubiorev.2018.11.011

    • PubMed
    • Google Scholar
37. 1. Hosford PS
    2. Wells JA
    3. Nizari S
    4. Christie IN
    5. Theparambil SM
    6. Castro PA
    7. Hadjihambi A
    8. Barros LF
    9. Ruminot I
    10. Lythgoe MF
    11. Gourine AV

    (2022) CO₂ signaling mediates neurovascular coupling in the cerebral cortex

    Nature Communications 13:2125.

    https://doi.org/10.1038/s41467-022-29622-9

    • PubMed
    • Google Scholar
38. 1. Ishai A
    2. Ungerleider LG
    3. Martin A
    4. Haxby JV

    (2000) The representation of objects in the human occipital and temporal cortex

    Journal of Cognitive Neuroscience 12 Suppl 2:35–51.

    https://doi.org/10.1162/089892900564055

    • PubMed
    • Google Scholar
39. Report

    1. Jenkinson M
    2. Pechaud M
    3. Smith S

    (2005)

    BET2: MR-based estimation of brain, skull and scalp surfaces

    Oxford university centre for functional MRI of the brain.

    • Google Scholar
40. 1. Jeong M
    2. Tashiro M
    3. Singh LN
    4. Yamaguchi K
    5. Horikawa E
    6. Miyake M
    7. Watanuki S
    8. Iwata R
    9. Fukuda H
    10. Takahashi Y
    11. Itoh M

    (2006) Functional brain mapping of actual car-driving using [18F]FDG-PET

    Annals of Nuclear Medicine 20:623–628.

    https://doi.org/10.1007/BF02984660

    • PubMed
    • Google Scholar
41. 1. Johnson MH

    (2001) Functional brain development in humans

    Nature Reviews. Neuroscience 2:475–483.

    https://doi.org/10.1038/35081509

    • PubMed
    • Google Scholar
42. 1. Johnson MH

    (2011) Interactive specialization: A domain-general framework for human functional brain development?

    Developmental Cognitive Neuroscience 1:7–21.

    https://doi.org/10.1016/j.dcn.2010.07.003

    • PubMed
    • Google Scholar
43. 1. Jones EJH
    2. Venema K
    3. Lowy R
    4. Earl RK
    5. Webb SJ

    (2015) Developmental changes in infant brain activity during naturalistic social experiences

    Developmental Psychobiology 57:842–853.

    https://doi.org/10.1002/dev.21336

    • PubMed
    • Google Scholar
44. 1. Kann O

    (2011) The energy demand of fast neuronal network oscillations: insights from brain slice preparations

    Frontiers in Pharmacology 2:90.

    https://doi.org/10.3389/fphar.2011.00090

    • PubMed
    • Google Scholar
45. 1. Kilner JM
    2. Mattout J
    3. Henson R
    4. Friston KJ

    (2005) Hemodynamic correlates of EEG: A heuristic

    NeuroImage 28:280–286.

    https://doi.org/10.1016/j.neuroimage.2005.06.008

    • PubMed
    • Google Scholar
46. 1. Koch SP
    2. Werner P
    3. Steinbrink J
    4. Fries P
    5. Obrig H

    (2009) Stimulus-induced and state-dependent sustained gamma activity is tightly coupled to the hemodynamic response in humans

    The Journal of Neuroscience 29:13962–13970.

    https://doi.org/10.1523/JNEUROSCI.1402-09.2009

    • Google Scholar
47. 1. Kosakowski HL
    2. Cohen MA
    3. Takahashi A
    4. Keil B
    5. Kanwisher N
    6. Saxe R

    (2022) Selective responses to faces, scenes, and bodies in the ventral visual pathway of infants

    Current Biology 32:265–274.

    https://doi.org/10.1016/j.cub.2021.10.064

    • PubMed
    • Google Scholar
48. 1. Kozberg M
    2. Hillman E

    (2016) Neurovascular coupling and energy metabolism in the developing brain

    Progress in Brain Research 225:213–242.

    https://doi.org/10.1016/bs.pbr.2016.02.002

    • PubMed
    • Google Scholar
49. 1. Kucewicz MT
    2. Cimbalnik J
    3. Matsumoto JY
    4. Brinkmann BH
    5. Bower MR
    6. Vasoli V
    7. Sulc V
    8. Meyer F
    9. Marsh WR
    10. Stead SM
    11. Worrell GA

    (2014) High frequency oscillations are associated with cognitive processing in human recognition memory

    Brain 137:2231–2244.

    https://doi.org/10.1093/brain/awu149

    • PubMed
    • Google Scholar
50. 1. Leech R
    2. Scott G
    3. Carhart-Harris R
    4. Turkheimer F
    5. Taylor-Robinson SD
    6. Sharp DJ

    (2014) Spatial dependencies between large-scale brain networks

    PLOS ONE 9:e98500.

    https://doi.org/10.1371/journal.pone.0098500

    • PubMed
    • Google Scholar
51. 1. Lloyd-Fox S
    2. Blasi A
    3. Volein A
    4. Everdell N
    5. Elwell CE
    6. Johnson MH

    (2009) Social perception in infancy: a near infrared spectroscopy study

    Child Development 80:986–999.

    https://doi.org/10.1111/j.1467-8624.2009.01312.x

    • PubMed
    • Google Scholar
52. 1. Lloyd-Fox S
    2. Blasi A
    3. Mercure E
    4. Elwell CE
    5. Johnson MH

    (2012) The emergence of cerebral specialization for the human voice over the first months of life

    Social Neuroscience 7:317–330.

    https://doi.org/10.1080/17470919.2011.614696

    • PubMed
    • Google Scholar
53. 1. Lloyd-Fox S
    2. Blasi A
    3. Elwell CE
    4. Charman T
    5. Murphy D
    6. Johnson MH

    (2013) Reduced neural sensitivity to social stimuli in infants at risk for autism

    Proceedings of the Royal Society B 280:20123026.

    https://doi.org/10.1098/rspb.2012.3026

    • Google Scholar
54. 1. Lloyd-Fox S
    2. Papademetriou M
    3. Darboe MK
    4. Everdell NL
    5. Wegmuller R
    6. Prentice AM
    7. Moore SE
    8. Elwell CE

    (2014) Functional near infrared spectroscopy (fNIRS) to assess cognitive function in infants in rural Africa

    Scientific Reports 4:4740.

    https://doi.org/10.1038/srep04740

    • PubMed
    • Google Scholar
55. 1. Lloyd-Fox S
    2. Begus K
    3. Halliday D
    4. Pirazzoli L
    5. Blasi A
    6. Papademetriou M
    7. Darboe MK
    8. Prentice AM
    9. Johnson MH
    10. Moore SE
    11. Elwell CE

    (2017) Cortical specialisation to social stimuli from the first days to the second year of life: A rural Gambian cohort

    Developmental Cognitive Neuroscience 25:92–104.

    https://doi.org/10.1016/j.dcn.2016.11.005

    • PubMed
    • Google Scholar
56. 1. Lloyd‐Fox S
    2. Blasi A
    3. Pasco G
    4. Gliga T
    5. Jones EJH
    6. Murphy DGM
    7. Elwell CE
    8. Charman T
    9. Johnson MH
    10. the BASIS Team

    (2018) Cortical responses before 6 months of life associate with later autism

    European Journal of Neuroscience 47:736–749.

    https://doi.org/10.1111/ejn.13757

    • Google Scholar
57. 1. Logothetis NK
    2. Pauls J
    3. Augath M
    4. Trinath T
    5. Oeltermann A

    (2001) Neurophysiological investigation of the basis of the fMRI signal

    Nature 412:150–157.

    https://doi.org/10.1038/35084005

    • PubMed
    • Google Scholar
58. 1. Lundgaard I
    2. Li B
    3. Xie L
    4. Kang H
    5. Sanggaard S
    6. Haswell JDR
    7. Sun W
    8. Goldman S
    9. Blekot S
    10. Nielsen M
    11. Takano T
    12. Deane R
    13. Nedergaard M

    (2015) Direct neuronal glucose uptake heralds activity-dependent increases in cerebral metabolism

    Nature Communications 6:6807.

    https://doi.org/10.1038/ncomms7807

    • PubMed
    • Google Scholar
59. 1. Malonek D
    2. Grinvald A

    (1996) Interactions between electrical activity and cortical microcirculation revealed by imaging spectroscopy: implications for functional brain mapping

    Science 272:551–554.

    https://doi.org/10.1126/science.272.5261.551

    • PubMed
    • Google Scholar
60. 1. Masina F
    2. Pezzetta R
    3. Lago S
    4. Mantini D
    5. Scarpazza C
    6. Arcara G

    (2022) Disconnection from prediction: A systematic review on the role of right temporoparietal junction in aberrant predictive processing

    Neuroscience and Biobehavioral Reviews 138:104713.

    https://doi.org/10.1016/j.neubiorev.2022.104713

    • PubMed
    • Google Scholar
61. 1. Minagawa-Kawai Y
    2. van der Lely H
    3. Ramus F
    4. Sato Y
    5. Mazuka R
    6. Dupoux E

    (2011) Optical brain imaging reveals general auditory and language-specific processing in early infant development

    Cerebral Cortex 21:254–261.

    https://doi.org/10.1093/cercor/bhq082

    • PubMed
    • Google Scholar
62. 1. Molavi B
    2. Dumont GA

    (2012) Wavelet-based motion artifact removal for functional near-infrared spectroscopy

    Physiological Measurement 33:259–270.

    https://doi.org/10.1088/0967-3334/33/2/259

    • PubMed
    • Google Scholar
63. 1. Müller-Bardorff M
    2. Bruchmann M
    3. Mothes-Lasch M
    4. Zwitserlood P
    5. Schlossmacher I
    6. Hofmann D
    7. Miltner W
    8. Straube T

    (2018) Early brain responses to affective faces: A simultaneous EEG-fMRI study

    NeuroImage 178:660–667.

    https://doi.org/10.1016/j.neuroimage.2018.05.081

    • PubMed
    • Google Scholar
64. 1. Nguyen VT
    2. Breakspear M
    3. Cunnington R

    (2014) Fusing concurrent EEG-fMRI with dynamic causal modeling: application to effective connectivity during face perception

    NeuroImage 102 Pt 1:60–70.

    https://doi.org/10.1016/j.neuroimage.2013.06.083

    • PubMed
    • Google Scholar
65. 1. Nguyen VT
    2. Cunnington R

    (2014) The superior temporal sulcus and the N170 during face processing: single trial analysis of concurrent EEG-fMRI

    NeuroImage 86:492–502.

    https://doi.org/10.1016/j.neuroimage.2013.10.047

    • PubMed
    • Google Scholar
66. 1. Niessing J
    2. Ebisch B
    3. Schmidt KE
    4. Niessing M
    5. Singer W
    6. Galuske RAW

    (2005) Hemodynamic signals correlate tightly with synchronized gamma oscillations

    Science 309:948–951.

    https://doi.org/10.1126/science.1110948

    • PubMed
    • Google Scholar
67. 1. Norman MG
    2. O’Kusky JR

    (1986) The growth and development of microvasculature in human cerebral cortex

    Journal of Neuropathology and Experimental Neurology 45:222–232.

    https://doi.org/10.1097/00005072-198605000-00004

    • PubMed
    • Google Scholar
68. 1. Oswald MC
    2. Brooks PS
    3. Zwart MF
    4. Mukherjee A
    5. West RJ
    6. Giachello CN
    7. Morarach K
    8. Baines RA
    9. Sweeney ST
    10. Landgraf M

    (2018) Reactive oxygen species regulate activity-dependent neuronal plasticity in Drosophila

    eLife 7:e39393.

    https://doi.org/10.7554/eLife.39393

    • PubMed
    • Google Scholar
69. 1. Pelphrey KA
    2. Morris JP
    3. McCarthy G

    (2005) Neural basis of eye gaze processing deficits in autism

    Brain 128:1038–1048.

    https://doi.org/10.1093/brain/awh404

    • PubMed
    • Google Scholar
70. 1. Phan P
    2. Highton D
    3. Lai J
    4. Smith M
    5. Elwell C
    6. Tachtsidis I

    (2016) Multi-channel multi-distance broadband near-infrared spectroscopy system to measure the spatial response of cellular oxygen metabolism and tissue oxygenation

    Biomedical Optics Express 7:4424–4440.

    https://doi.org/10.1364/BOE.7.004424

    • PubMed
    • Google Scholar
71. 1. Pinti P
    2. Siddiqui MF
    3. Levy AD
    4. Jones EJH
    5. Tachtsidis I

    (2021) An analysis framework for the integration of broadband NIRS and EEG to assess neurovascular and neurometabolic coupling

    Scientific Reports 11:3977.

    https://doi.org/10.1038/s41598-021-83420-9

    • PubMed
    • Google Scholar
72. 1. Raichle ME
    2. Mintun MA

    (2006) Brain work and brain imaging

    Annual Review of Neuroscience 29:449–476.

    https://doi.org/10.1146/annurev.neuro.29.051605.112819

    • PubMed
    • Google Scholar
73. 1. Rocher AB
    2. Chapon F
    3. Blaizot X
    4. Baron JC
    5. Chavoix C

    (2003) Resting-state brain glucose utilization as measured by PET is directly related to regional synaptophysin levels: a study in baboons

    NeuroImage 20:1894–1898.

    https://doi.org/10.1016/j.neuroimage.2003.07.002

    • PubMed
    • Google Scholar
74. 1. Rosa MJ
    2. Kilner J
    3. Blankenburg F
    4. Josephs O
    5. Penny W

    (2010) Estimating the transfer function from neuronal activity to BOLD using simultaneous EEG-fMRI

    NeuroImage 49:1496–1509.

    https://doi.org/10.1016/j.neuroimage.2009.09.011

    • PubMed
    • Google Scholar
75. 1. Rowan RA
    2. Maxwell DS

    (1981) Patterns of vascular sprouting in the postnatal development of the cerebral cortex of the rat

    The American Journal of Anatomy 160:247–255.

    https://doi.org/10.1002/aja.1001600303

    • PubMed
    • Google Scholar
76. 1. Sato W
    2. Kochiyama T
    3. Uono S
    4. Matsuda K
    5. Usui K
    6. Inoue Y
    7. Toichi M

    (2014) Rapid, high-frequency, and theta-coupled gamma oscillations in the inferior occipital gyrus during face processing

    Cortex; a Journal Devoted to the Study of the Nervous System and Behavior 60:52–68.

    https://doi.org/10.1016/j.cortex.2014.02.024

    • PubMed
    • Google Scholar
77. 1. Scheeringa R
    2. Petersson KM
    3. Oostenveld R
    4. Norris DG
    5. Hagoort P
    6. Bastiaansen MCM

    (2009) Trial-by-trial coupling between EEG and BOLD identifies networks related to alpha and theta EEG power increases during working memory maintenance

    NeuroImage 44:1224–1238.

    https://doi.org/10.1016/j.neuroimage.2008.08.041

    • PubMed
    • Google Scholar
78. 1. Scheeringa R
    2. Fries P
    3. Petersson K-M
    4. Oostenveld R
    5. Grothe I
    6. Norris DG
    7. Hagoort P
    8. Bastiaansen MCM

    (2011) Neuronal dynamics underlying high- and low-frequency EEG oscillations contribute independently to the human BOLD signal

    Neuron 69:572–583.

    https://doi.org/10.1016/j.neuron.2010.11.044

    • Google Scholar
79. 1. Scheeringa R
    2. Petersson KM
    3. Kleinschmidt A
    4. Jensen O
    5. Bastiaansen MCM

    (2012) EEG α power modulation of fMRI resting-state connectivity

    Brain Connectivity 2:254–264.

    https://doi.org/10.1089/brain.2012.0088

    • PubMed
    • Google Scholar
80. 1. Schroeter ML
    2. Bücheler MM
    3. Müller K
    4. Uludağ K
    5. Obrig H
    6. Lohmann G
    7. Tittgemeyer M
    8. Villringer A
    9. von Cramon DY

    (2004) Towards a standard analysis for functional near-infrared imaging

    NeuroImage 21:283–290.

    https://doi.org/10.1016/j.neuroimage.2003.09.054

    • PubMed
    • Google Scholar
81. 1. Schurz M
    2. Radua J
    3. Aichhorn M
    4. Richlan F
    5. Perner J

    (2014) Fractionating theory of mind: a meta-analysis of functional brain imaging studies

    Neuroscience and Biobehavioral Reviews 42:9–34.

    https://doi.org/10.1016/j.neubiorev.2014.01.009

    • PubMed
    • Google Scholar
82. 1. Schurz M
    2. Tholen MG
    3. Perner J
    4. Mars RB
    5. Sallet J

    (2017) Specifying the brain anatomy underlying temporo-parietal junction activations for theory of mind: A review using probabilistic atlases from different imaging modalities

    Human Brain Mapping 38:4788–4805.

    https://doi.org/10.1002/hbm.23675

    • PubMed
    • Google Scholar
83. 1. Schweiger M
    2. Arridge S

    (2014) The Toast++ software suite for forward and inverse modeling in optical tomography

    Journal of Biomedical Optics 19:040801.

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

    • PubMed
    • Google Scholar
84. 1. Seregi A
    2. Keller M
    3. Hertting G

    (1987) Are cerebral prostanoids of astroglial origin? Studies on the prostanoid forming system in developing rat brain and primary cultures of rat astrocytes

    Brain Research 404:113–120.

    https://doi.org/10.1016/0006-8993(87)91361-8

    • PubMed
    • Google Scholar
85. 1. Shi F
    2. Yap PT
    3. Wu G
    4. Jia H
    5. Gilmore JH
    6. Lin W
    7. Shen D

    (2011) Infant brain atlases from neonates to 1- and 2-year-olds

    PLOS ONE 6:e18746.

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

    • PubMed
    • Google Scholar
86. 1. Shimada S
    2. Hiraki K

    (2006) Infant’s brain responses to live and televised action

    NeuroImage 32:930–939.

    https://doi.org/10.1016/j.neuroimage.2006.03.044

    • PubMed
    • Google Scholar
87. 1. Shokri-Kojori E
    2. Tomasi D
    3. Alipanahi B
    4. Wiers CE
    5. Wang GJ
    6. Volkow ND

    (2019) Correspondence between cerebral glucose metabolism and BOLD reveals relative power and cost in human brain

    Nature Communications 10:690.

    https://doi.org/10.1038/s41467-019-08546-x

    • PubMed
    • Google Scholar
88. 1. Siddiqui MF
    2. Elwell CE
    3. Johnson MH

    (2016) Mitochondrial dysfunction in autism spectrum disorders

    Autism-Open Access 6:1000190.

    https://doi.org/10.4172/2165-7890.1000190

    • PubMed
    • Google Scholar
89. 1. Siddiqui MF
    2. Pinti P
    3. Lloyd-Fox S
    4. Jones EJH
    5. Brigadoi S
    6. Collins-Jones L
    7. Tachtsidis I
    8. Johnson MH
    9. Elwell CE

    (2021) Regional haemodynamic and metabolic coupling in infants

    Frontiers in Human Neuroscience 15:780076.

    https://doi.org/10.3389/fnhum.2021.780076

    • PubMed
    • Google Scholar
90. Software

    1. Siddiqui M

    (2023) Nirs-EEG, version swh:1:rev:1af4d1f0b4699f9491f43443634a2cf2ba61b3b6

    Software Heritage.

    https://archive.softwareheritage.org/swh:1:dir:2a6b8853e7b5a32be4702ca6924b4d38cfdd03ce;origin=https://github.com/maheensiddiqui91/NIRS-EEG;visit=swh:1:snp:c40ff99dec947290126d9a4c42d6f0bdbb76f1d7;anchor=swh:1:rev:1af4d1f0b4699f9491f43443634a2cf2ba61b3b6
91. 1. Smith AJ
    2. Blumenfeld H
    3. Behar KL
    4. Rothman DL
    5. Shulman RG
    6. Hyder F

    (2002) Cerebral energetics and spiking frequency: the neurophysiological basis of fMRI

    PNAS 99:10765–10770.

    https://doi.org/10.1073/pnas.132272199

    • PubMed
    • Google Scholar
92. 1. Sokoloff L

    (1999) Energetics of functional activation in neural tissues

    Neurochemical Research 24:321–329.

    https://doi.org/10.1023/a:1022534709672

    • PubMed
    • Google Scholar
93. 1. Steiner P

    (2019) Brain fuel utilization in the developing brain

    Annals of Nutrition & Metabolism 75 Suppl 1:8–18.

    https://doi.org/10.1159/000508054

    • PubMed
    • Google Scholar
94. 1. Stichel CC
    2. Müller CM
    3. Zilles K

    (1991) Distribution of glial fibrillary acidic protein and vimentin immunoreactivity during rat visual cortex development

    Journal of Neurocytology 20:97–108.

    https://doi.org/10.1007/BF01279614

    • PubMed
    • Google Scholar
95. 1. Strangman G
    2. Culver JP
    3. Thompson JH
    4. Boas DA

    (2002) A quantitative comparison of simultaneous BOLD fMRI and NIRS recordings during functional brain activation

    NeuroImage 17:719–731.

    https://doi.org/10.1016/S1053-8119(02)91227-9

    • PubMed
    • Google Scholar
96. 1. Uono S
    2. Sato W
    3. Kochiyama T
    4. Kubota Y
    5. Sawada R
    6. Yoshimura S
    7. Toichi M

    (2017) Time course of gamma-band oscillation associated with face processing in the inferior occipital gyrus and fusiform gyrus: A combined fMRI and MEG study

    Human Brain Mapping 38:2067–2079.

    https://doi.org/10.1002/hbm.23505

    • PubMed
    • Google Scholar
97. 1. Vaishnavi SN
    2. Vlassenko AG
    3. Rundle MM
    4. Snyder AZ
    5. Mintun MA
    6. Raichle ME

    (2010) Regional aerobic glycolysis in the human brain

    PNAS 107:17757–17762.

    https://doi.org/10.1073/pnas.1010459107

    • PubMed
    • Google Scholar
98. 1. van der Velde B
    2. White T
    3. Kemner C

    (2021) The emergence of a theta social brain network during infancy

    NeuroImage 240:118298.

    https://doi.org/10.1016/j.neuroimage.2021.118298

    • PubMed
    • Google Scholar
99. 1. Vaynman S
    2. Ying Z
    3. Wu A
    4. Gomez-Pinilla F

    (2006) Coupling energy metabolism with a mechanism to support brain-derived neurotrophic factor-mediated synaptic plasticity

    Neuroscience 139:1221–1234.

    https://doi.org/10.1016/j.neuroscience.2006.01.062

    • PubMed
    • Google Scholar
100. 1. Vezyroglou A
     2. Hebden P
     3. De Roever I
     4. Thornton R
     5. Mitra S
     6. Worley A
     7. Alves M
     8. Dean E
     9. Cross JH
     10. Tachtsidis I

     (2022) Broadband-NIRS system identifies epileptic focus in a child with focal cortical dysplasia-a case study

     Metabolites 12:260.

     https://doi.org/10.3390/metabo12030260

     • PubMed
     • Google Scholar
101. 1. von Stein A
     2. Sarnthein J

     (2000) Different frequencies for different scales of cortical integration: from local gamma to long range alpha/theta synchronization

     International Journal of Psychophysiology 38:301–313.

     https://doi.org/10.1016/s0167-8760(00)00172-0

     • PubMed
     • Google Scholar
102. Preprint

     1. Wang Y
     2. Royer J
     3. Park B
     4. de Wael RV
     5. Larivière S
     6. Tavakol S
     7. Rodriguez-Cruces R
     8. Paquola C
     9. Hong SJ
     10. Margulies DS
     11. Smallwood J
     12. Valk SL
     13. Evans AC
     14. Bernhardt BC

     (2021) Long-Range Connections Mirror and Link Microarchitectural and Cognitive Hierarchies in the Human Brain

     bioRxiv.

     https://doi.org/10.1101/2021.10.25.465692

     • Google Scholar
103. 1. Wilcox T
     2. Haslup JA
     3. Boas DA

     (2010) Dissociation of processing of featural and spatiotemporal information in the infant cortex

     NeuroImage 53:1256–1263.

     https://doi.org/10.1016/j.neuroimage.2010.06.064

     • PubMed
     • Google Scholar
104. 1. Wilterson AI
     2. Nastase SA
     3. Bio BJ
     4. Guterstam A
     5. Graziano MSA

     (2021) Attention, awareness, and the right temporoparietal junction

     PNAS 118:e2026099118.

     https://doi.org/10.1073/pnas.2026099118

     • PubMed
     • Google Scholar
105. 1. Wong-Riley MT
     2. Hevner RF
     3. Cutlan R
     4. Earnest M
     5. Egan R
     6. Frost J
     7. Nguyen T

     (1993) Cytochrome oxidase in the human visual cortex: distribution in the developing and the adult brain

     Visual Neuroscience 10:41–58.

     https://doi.org/10.1017/s0952523800003217

     • PubMed
     • Google Scholar
106. 1. Yuan H
     2. Liu T
     3. Szarkowski R
     4. Rios C
     5. Ashe J
     6. He B

     (2010) Negative covariation between task-related responses in alpha/beta-band activity and BOLD in human sensorimotor cortex: an EEG and fMRI study of motor imagery and movements

     NeuroImage 49:2596–2606.

     https://doi.org/10.1016/j.neuroimage.2009.10.028

     • PubMed
     • Google Scholar
107. 1. Zhao J
     2. Ding HS
     3. Hou XL
     4. Zhou CL
     5. Chance B

     (2005) In vivo determination of the optical properties of infant brain using frequency-domain near-infrared spectroscopy

     Journal of Biomedical Optics 10:024028.

     https://doi.org/10.1117/1.1891345

     • PubMed
     • Google Scholar
108. 1. Zion-Golumbic E
     2. Golan T
     3. Anaki D
     4. Bentin S

     (2008) Human face preference in gamma-frequency EEG activity

     NeuroImage 39:1980–1987.

     https://doi.org/10.1016/j.neuroimage.2007.10.025

     • PubMed
     • Google Scholar

Decision letter after peer review:

Thank you for submitting your article "Mapping human social brain specialisation beyond the neuron using multimodal imaging in human infants" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Chris Baker as the Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: Claire Kabdebon (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

Although the reviewers acknowledged the important methodological contribution of your study, the following key aspects of the article were questioned and will need to be addressed in the revised manuscript, in addition to the other reviewers' comments (detailed below).

1) A discrepancy was perceived between the main claim of the article at the cognitive/developmental level (on brain specialization for social processing in infants) and the evidence provided by this work. It seems necessary to temper this presentation because the analyses and results are not as aligned with the aims and hypotheses as the reader might expect after reading the abstract and introduction. To gain clarity and impact of the manuscript, we advise you to partially reframe the article to better highlight the methodological contribution rather than emphasizing the fundamental side which is insufficiently supported by the results.

2) At the methodological level, avoiding shortcuts and general statements would be welcome. Better specification of your predictions in complex analyses such as the final general linear model would facilitate the interpretation of the results. The spatial comparison of source-localized NIRS measures and scalp-based EEG measures was also questioned.

3) Finally, during our discussions following the individual review process, Reviewer #2 suggested adding a critical discussion of the difference in hemodynamic response shape when signals are grand averaged vs. the HRF shape.

Reviewer #1 (Recommendations for the authors):

I feel the manuscript would be stronger if it focussed on evaluating the combination of methods. At the moment, it takes on too much and so ultimately disappoints. Theoretically, the introduction builds a case that this method might provide a special test of interactive specialization, but this is not revisited in the results or the discussion. And methodologically, there are large claims throughout (e.g., the title "beyond the neuron", line 73 "work with single modalities…does not yield insights", line 217 "a critical nexus in understanding localisation", line 254 "novel multimodal and non-invasive", line 268 "demonstrates the strength… particularly in clinical populations"). These are non-substantiated and distract from the genuine discoveries in the paper. Highly relevant literature is also omitted. For example, there is a whole community using EEG and fMRI, either sequentially or simultaneously, that has produced what I expect are hundreds of papers. What lessons have been learned from this? A summary should be given in the introduction and the work-related back in the discussion.

The authors perform source localisation to map the scalp measures to brain sources for the NIRS. It would be interesting to give some estimate of the spatial resolution of this method. But no such localisation is attempted for EEG. It was probably omitted because there are too few EEG channels for it to be meaningful. But, it is then just not possible to spatially compare scalp-based EEG measures, which may be far from the neural source to a neural measure, of localised NIRS. I feel that plotting the scalp electrodes on the brain, along with the NIRS data, is misleading.

It is difficult to read a paper where the methods are so important, with the methods at the end.

Reviewer #2 (Recommendations for the authors):

Please check two issues with Graphic links in the text: lines 374 and 561 "Error! Reference source not found."

Reviewer #3 (Recommendations for the authors):

Here are some additional remarks:

– In order to address the specialization process, and if the sample size and distribution allow, it might be interesting to explore age-related effects over the reported brain regions.

– About the organization of the methods section, for clarification, I would suggest following the structure of the Results section.

– Regarding the EEG analysis, some aspects of the analysis were unclear to me. First, about data segmentation: it is mentioned that data is epoched in 1s segments, consisting of 200ms of the previous segment and 800ms of the current segment: it is thus my understanding that segments are overlapping. However, it doesn't fit with the schematic representation of the procedure in figure 6b. This procedure should be clarified.

Second, it is also mentioned that the first 200ms of the segment was used as a baseline. What is the purpose of this baseline correction, since a second baseline correction will be performed in the frequency domain? Is it only for artifact detection purposes?

Third, regarding the definition of the frequency bands, the β and γ bands are overlapping (Β from 13 to 30Hz and Γ from 20 to 60Hz), which means that the two measures are not independent. How do the authors ensure that the reported effects in the two frequency bands are not driven by this shared activity?

– About the hemodynamic to metabolic coupling assessment, it is mentioned that the coupling of hemodynamic and metabolic activity was restricted to specific channels, based on previous studies. Which are these specific channels?

– Regarding the HRF and MRF reconstruction, the schematic representation of the procedure in figure 7 was confusing to me. The authors performed block averages of the HbO, HbR, and COO signals. However, it is unclear to me whether the blocks consist of [baseline stimulation] or [stimulation baseline]: what does the 0-time point correspond to in these figures? It seems like on the group average response on the top it represents the onset of the baseline, and on the reconstructed infant response functions it represents the onset of the stimulation. This is only a detail of the representation, but it was confusing for me.

Besides, do the depicted response functions represent the actual functions obtained? It would be informative to provide the reader with this information.

– Regarding the combined NIRS-EEG analysis, it was unclear to me whether the EEG measures were used to model the bNIRS data on a second-by-second basis (i.e. the time-resolved rms power signal is convolved with the response function), or on a block by block basis (i.e. average of the rms power signal for each block), or whether an average across each of the two conditions was used (i.e. average rms power across all blocks of each condition). Besides, L463-465 it is mentioned that the RMS power is entered in the GLM with 2s baseline + 8s stim, but L545, it is then mentioned that only the data for the stimulus period is included. Please clarify this point.

– Regarding multiple comparisons, the proposed analysis examining relations between EEG activity in 4 frequency bands to each of the three chromophores implies a high number of comparisons. The authors indicate that they used FDR corrections: across which dimensions were these corrections performed? Is it EEG x bNIRS sensors x chromophores x frequency bands? It would be worth mentioning for each of the analyses performed.

– Please mention the average amount of data used to perform each of the analyses across infants (i.e. how many blocks of stimulation are included).

– I found that some of the information included in the figures was sometimes too small or very hard to read. I would suggest re-designing some of these figures to be self-explanatory.

– Figure 2-f: it is unclear what this graph corresponds to, and how it was obtained. Does it correspond to a specific condition social or non-social? What is the added value of this graph in the framework of this study?

– There are some reference errors (L561 and 374).

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Using multi-modal neuroimaging to characterise social brain specialisation in infants" for further consideration by eLife. Your revised article has been evaluated by Floris de Lange (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below.

Reviewer #3 (Recommendations for the authors):

In general, the authors produced a stronger manuscript that addresses most of the concerns that were raised. In particular, the authors improved the overall clarity of their work both conceptually and methodologically, and they provide better data visualization. Besides, the scope and contribution of the paper is now communicated more clearly.

Yet, a few aspects of the manuscript remain a bit obscure to me and I believe they should be further improved or at least addressed.

First, after reading the paper, the pattern of results obtained from the combined EEG – bNIRS analysis does not seem as clear as I would have expected from the authors predictions. In figures 3-1 and 3-2, although some bNIRS channels (11 and 14) are overall more consistently reported than others as significant, as emphasized in the results and discussion, there is still quite some variability in the pattern of bNIRS to EEG channels coupling across chromophores (and frequency bands and conditions). How do the authors explain this variability? Shouldn't the different chromophore measurements reflect a similar phenomenon, and thus display similar coupling patterns with EEG channels? Is this the expected level of variability for this kind of dataset, or for this age range, population size etc….? It would be really informative if the authors could elaborate more on the reasons why we observe this variability in the discussion, and if possible provide recommendations on how to reduce it in further studies.

Second, about the authors' predictions in the introduction, they mention that they expect to observe "coordinated increases in hemodynamics/metabolism and neural oscillatory activity" in the social condition, and "coordinated decreases in hemodynamics/metabolism and neural oscillatory activity" in the non-social condition. This wording suggests that this is a hypothesis that the authors will be able to test. However, the authors do not test this hypothesis further in the manuscript and it seems to me that the directionality of coordinated activity cannot be directly assessed using this GLM approach. I would therefore suggest to slightly reframe the phrasing of this paragraph to avoid that confusion.

Essential revisions:

1) A discrepancy was perceived between the main claim of the article at the cognitive/developmental level (on brain specialization for social processing in infants) and the evidence provided by this work. It seems necessary to temper this presentation because the analyses and results are not as aligned with the aims and hypotheses as the reader might expect after reading the abstract and introduction. To gain clarity and impact of the manuscript, we advise you to partially reframe the article to better highlight the methodological contribution rather than emphasizing the fundamental side which is insufficiently supported by the results.

We would like to thank the editors for their comments. Pertaining to the first point, the introduction and Discussion sections of the article have been reframed to emphasize the methodological impact. We have also changed the title to reflect an increased methodological focus. We have retained the theoretical background that motivates the development of the measures but made it clearer that the primary contribution of this article is in developing these methods.

2) At the methodological level, avoiding shortcuts and general statements would be welcome. Better specification of your predictions in complex analyses such as the final general linear model would facilitate the interpretation of the results. The spatial comparison of source-localized NIRS measures and scalp-based EEG measures was also questioned.

We have removed any general statements and have included hypotheses to facilitate interpretation of results. Further, to clarify, we did not perform any source-localisation on the NIRS data and therefore we did not compare scalp-based EEG and source-localised NIRS. We have added text in the article to ensure this is clear.

3) Finally, during our discussions following the individual review process, Reviewer #2 suggested adding a critical discussion of the difference in hemodynamic response shape when signals are grand averaged vs. the HRF shape.

Regarding the last point, we have clarified both in the text (Line 1554-1555) and in Figure 7 that the estimated HRF and MRF correspond to the basis function to be used within the GLM (i.e. analogous to the canonical HRF typically used for fMRI and fNIRS GLM-based analyses and convolved with boxcar functions to generate regressors). These hence correspond to the hemodynamic/metabolic response to an event of 0 duration and differ from the group averaged responses that represent the responses averaged across the whole duration of the various task blocks. The HRF and MRF were estimated using the iterative procedure as these are not known for infants and the MRF nor for infants and adults, therefore providing a new method to optimize the analysis of NIRS-based data when the canonical HRF might not be appropriate.

Reviewer #1 (Recommendations for the authors):

I feel the manuscript would be stronger if it focussed on evaluating the combination of methods. At the moment, it takes on too much and so ultimately disappoints. Theoretically, the introduction builds a case that this method might provide a special test of interactive specialization, but this is not revisited in the results or the discussion. And methodologically, there are large claims throughout (e.g., the title "beyond the neuron", line 73 "work with single modalities…does not yield insights", line 217 "a critical nexus in understanding localisation", line 254 "novel multimodal and non-invasive", line 268 "demonstrates the strength… particularly in clinical populations"). These are non-substantiated and distract from the genuine discoveries in the paper. Highly relevant literature is also omitted. For example, there is a whole community using EEG and fMRI, either sequentially or simultaneously, that has produced what I expect are hundreds of papers. What lessons have been learned from this? A summary should be given in the introduction and the work-related back in the discussion.

We thank the reviewer for their comment. We revised the introduction to increase the focus on the methodological advances represented by our work, though we think the broader theoretical context that motivates the use of these methods in general is still important to consider (because that provides the motivation for developing them). We have clarified that we are suggesting that integrating metabolic measurements and their coupling to other measures may provide a more sensitive way of measuring functional specialisation (the selective engagement of a particular brain region in a particular process) because energetic constraints may play a role in interactive specialisation, one theory of how functional specialisation arises. We are not arguing that our measures can measure interactive specialisation itself, and so we have tried to highlight this distinction throughout. We have further clarified that this is the context in which we present material on theoretical frameworks and returned to them in the discussion from this perspective. We have also clarified several places where we used unclear phrasing (e.g., on line 1148 we meant that the study we were discussing on epilepsy had implications for use in clinical populations, rather than our study directly; we have clarified this now). We have added material on simultaneous EEG and fMRI to the introduction and discussion.

The authors perform source localisation to map the scalp measures to brain sources for the NIRS. It would be interesting to give some estimate of the spatial resolution of this method. But no such localisation is attempted for EEG. It was probably omitted because there are too few EEG channels for it to be meaningful. But, it is then just not possible to spatially compare scalp-based EEG measures, which may be far from the neural source to a neural measure, of localised NIRS. I feel that plotting the scalp electrodes on the brain, along with the NIRS data, is misleading.

We thank the reviewer for their comment. To clarify, we did not perform source localisation on the NIRS data for our analyses examining associations with EEG. We did perform image reconstruction (as seen in Figure 4) with the bNIRS data, but this was to address a separate goal. The HbO₂, HHb and oxCCO time-series (i.e. for each channel) were used as part of the combined bNIRS – EEG analysis, not the reconstructed time-series (i.e. for each node of the mesh). We have added text on lines 1623 – 1624 to make it clear that image reconstruction was a separate analysis that was performed with the main goal of visualizing how the changes in HbO₂, HHb, oxCCO were spatially related to each other. Moreover, we have changed Figure 6c to indicate at which step of the analysis image reconstruction was performed. This should also clarify that the data from the reconstruction was not used in any other subsequent analyses.

It is difficult to read a paper where the methods are so important, with the methods at the end.

We have followed the eLife template for this.

Reviewer #2 (Recommendations for the authors):

Please check two issues with Graphic links in the text: lines 374 and 561 "Error! Reference source not found."

This has been fixed.

Reviewer #3 (Recommendations for the authors):

– In order to address the specialization process, and if the sample size and distribution allow, it might be interesting to explore age-related effects over the reported brain regions.

We thank the reviewer for their suggestion which is indeed interesting and will be considered for future studies. Due to sample size limitations, we do not believe that age-related effects can be explored within the current dataset. However, our methods provide protocols for investigators to address this question in future studies.

– About the organization of the methods section, for clarification, I would suggest following the structure of the Results section.

We thank the reviewer for their suggestion. In its current form, the EEG results are presented first in the Results section, followed by the combined bNIRS – EEG results. The bNIRS results are not shown as they have already been published in a previous publication. Moreover, in the methods section, the bNIRS methodology is detailed first, followed by the EEG and then the combined bNIRS – EEG. This follows the same structure as the Results section.

– Regarding the EEG analysis, some aspects of the analysis were unclear to me. First, about data segmentation: it is mentioned that data is epoched in 1s segments, consisting of 200ms of the previous segment and 800ms of the current segment: it is thus my understanding that segments are overlapping. However, it doesn't fit with the schematic representation of the procedure in figure 6b. This procedure should be clarified.

The reviewer is correct in their understanding that the segments are overlapping. This has not been shown in Figure 6b as it would make the figure unnecessarily cluttered and difficult to understand. The point about overlapping segments has been emphasised in the caption of the figure.

Second, it is also mentioned that the first 200ms of the segment was used as a baseline. What is the purpose of this baseline correction, since a second baseline correction will be performed in the frequency domain? Is it only for artifact detection purposes?

It is standard procedure in EEG data analysis to perform baseline correction to get rid of drifts in the data and for the purposes of artifact detection.

Third, regarding the definition of the frequency bands, the β and γ bands are overlapping (Β from 13 to 30Hz and Γ from 20 to 60Hz), which means that the two measures are not independent. How do the authors ensure that the reported effects in the two frequency bands are not driven by this shared activity?

The frequency bands that are used are based on previous work by Saby and Marshall (2012) where infant frequency bands have been defined. These bands are widely adopted by EEG infant studies. In order to clear any doubts, we have now included a “high γ” frequency band (60 – 80 Hz) in the combined bNIRS – EEG analysis. The results shown in Figures 3b and 3c show consistency/similarity in the pattern of results in the “γ” and “high γ” frequency bands and no similarity between the “β” and “γ” frequency bands. We are therefore confident that the activity between the two bands is independent.

– About the hemodynamic to metabolic coupling assessment, it is mentioned that the coupling of hemodynamic and metabolic activity was restricted to specific channels, based on previous studies. Which are these specific channels?

This has been mentioned on line 1505 in the manuscript. To clarify, for each condition, for each chromophore, a different set of channels showed significant haemodynamic and metabolic coupling. To avoid further complexity, we decided to include all the channels that showed significant coupling for either of the conditions, for any of the chromophores. Therefore, considering all the significant channels across conditions and chromophores, the final channels included were 11, 12, 13, 14, 16 and 18.

– Regarding the HRF and MRF reconstruction, the schematic representation of the procedure in figure 7 was confusing to me. The authors performed block averages of the HbO, HbR, and COO signals. However, it is unclear to me whether the blocks consist of [baseline stimulation] or [stimulation baseline]: what does the 0-time point correspond to in these figures? It seems like on the group average response on the top it represents the onset of the baseline, and on the reconstructed infant response functions it represents the onset of the stimulation. This is only a detail of the representation, but it was confusing for me.

Besides, do the depicted response functions represent the actual functions obtained? It would be informative to provide the reader with this information.

We thank the reviewer for the comment. The block in the group-averaged responses corresponds to the period [stimulation + baseline] as infants often show delayed responses (e.g. see Lloyd-Fox et al., 2017 using a similar task). The HRF and MRF on the right of the figure instead represent the basis function, i.e. the model of the response to an event of 0 duration. In addition, the figure shows the actual functions obtained. In order to clarify these aspects, we have included the yellow shaded areas to indicate the stimulation periods as well as mentioned that the shown HRF and MRF are estimated functions.

– Regarding the combined NIRS-EEG analysis, it was unclear to me whether the EEG measures were used to model the bNIRS data on a second-by-second basis (i.e. the time-resolved rms power signal is convolved with the response function), or on a block by block basis (i.e. average of the rms power signal for each block), or whether an average across each of the two conditions was used (i.e. average rms power across all blocks of each condition). Besides, L463-465 it is mentioned that the RMS power is entered in the GLM with 2s baseline + 8s stim, but L545, it is then mentioned that only the data for the stimulus period is included. Please clarify this point.

We thank the reviewer for their comment. To clarify, the average RMS power (which is referred to as the “block averaged EEG response” in the text) across all blocks of each condition was used for the convolution with reconstructed infant HRF/MRF. This is because unlike fNIRS/bNIRS data, several trials of EEG data are required to accurately detect activity. It was therefore more accurate to average all the blocks of each condition for each participant. This has already been mentioned on line 1350 – “The blocks were then averaged across trials to obtain an averaged RMS response per participant” Regarding the reviewer’s second comment, there was indeed a discrepancy, and we had erroneously stated that the RMS power block that was entered into the GLM consisted of 2s baseline + 8s stimulus. In fact, only the stimulus period was used for the GLM analysis. This error has been corrected now on line 1352.

– Regarding multiple comparisons, the proposed analysis examining relations between EEG activity in 4 frequency bands to each of the three chromophores implies a high number of comparisons. The authors indicate that they used FDR corrections: across which dimensions were these corrections performed? Is it EEG x bNIRS sensors x chromophores x frequency bands? It would be worth mentioning for each of the analyses performed.

For the EEG analysis, the FDR correction was performed within each frequency band, across EEG channels. For the combined bNIRS – EEG analysis, the FDR correction was performed across EEG channels x bNIRS channels. This has been added to line 669 of the manuscript.

– Please mention the average amount of data used to perform each of the analyses across infants (i.e. how many blocks of stimulation are included).

6 blocks are used per infant on average for the bNIRS data. For the EEG data, an average of 30 1-s segments were used. This has been mentioned on line 1307 and 1326.

– I found that some of the information included in the figures was sometimes too small or very hard to read. I would suggest re-designing some of these figures to be self-explanatory.

We have tried to re-design the figures as and where possible. For example, for Figure 3, where the combined bNIRS – EEG results are presented, we have put the social and non-social results on one figure with two colour scales for ease of comparison. We have also stated explicitly what each colour represents to make it easier to interpret. Due to the number of results (bNIRS channels, EEG electrodes, frequency band and chromophores) it is difficult to make the results figures very big.

– Figure 2-f: it is unclear what this graph corresponds to, and how it was obtained. Does it correspond to a specific condition social or non-social? What is the added value of this graph in the framework of this study?

We thank the reviewer for their comment and have removed Figures 2e and 2f as they did not add any value to the manuscript.

– There are some reference errors (L561 and 374).

These have been checked.

[Editors’ note: what follows is the authors’ response to the second round of review.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below.

Reviewer #3 (Recommendations for the authors):

In general, the authors produced a stronger manuscript that addresses most of the concerns that were raised. In particular, the authors improved the overall clarity of their work both conceptually and methodologically, and they provide better data visualization. Besides, the scope and contribution of the paper is now communicated more clearly.

Yet, a few aspects of the manuscript remain a bit obscure to me and I believe they should be further improved or at least addressed.

First, after reading the paper, the pattern of results obtained from the combined EEG – bNIRS analysis does not seem as clear as I would have expected from the authors predictions. In figures 3-1 and 3-2, although some bNIRS channels (11 and 14) are overall more consistently reported than others as significant, as emphasized in the results and discussion, there is still quite some variability in the pattern of bNIRS to EEG channels coupling across chromophores (and frequency bands and conditions). How do the authors explain this variability? Shouldn't the different chromophore measurements reflect a similar phenomenon, and thus display similar coupling patterns with EEG channels? Is this the expected level of variability for this kind of dataset, or for this age range, population size etc….? It would be really informative if the authors could elaborate more on the reasons why we observe this variability in the discussion, and if possible provide recommendations on how to reduce it in further studies.

We thank the reviewer for their comment. We note that the reviewer also made a comment for us to clarify the FDR correction procedure and upon checking this (in order to include the exact details in the manuscript), we noted that the results figures that we had included after the previous revisions were not the FDR-corrected ones. The correct ones have now been included in Figure 3 supplement 1 and supplement 2 and are now much clearer. We see a more consistent, stimulus-dependent, localised pattern of associations between bNIRS channels and EEG electrodes. There is still a small amount of variation across chromophores, and we have added the following into the discussion on lines 340 – 395 which we hope will clarify why there is variability across chromophores and frequency bands:

“The strongest coupling was observed between temporo-parietal bNIRS channel 14 with parietal EEG locations Pz and PO4 for the social condition (for β and γ frequency bands). Meanwhile, for the non-social condition, coupling was observed between temporo-parietal bNIRS channel 14 with occipital and parietal EEG locations Oz, O2, PO8 and P10 (for theta and β frequency bands). While an overall consistent pattern of associations across chromophores and conditions was observed, some variability was also seen, particularly across frequency bands. This was expected and in line with previous EEG-fMRI studies that have demonstrated task-dependent variation in coupling between neural and haemodynamic activity across frequency bands [1]–[7]. For example, for resting state simultaneous fMRI and EEG, stronger coupling between the BOLD response and neural activity has been observed for the α band [8]. Meanwhile, for cognitive tasks, stronger coupling has been observed in the γ frequency band [9]. Scheeringa et al. (2011) investigated trial-by-trial coupling of EEG and BOLD activity and found that low- and high-frequency bands independently contribute to explaining BOLD variance. We therefore expected the frequency band showing the strongest coupling between bNIRS and EEG for each of the stimuli to vary. Further, while we did expect and observe significant overlap in associations between chromophores within each frequency band, some variability was seen. For example, for the social condition, no associations were observed in the low-frequency bands for any of the chromophores. Moreover, in the β frequency bands, all chromophores displayed significant associations between bNIRS channel 14 and Pz for the social condition and both HHb and oxCCO displayed significant associations between bNIRS channel 14 and O2, PO8 and C2. Similarly, in the γ frequency bands, both HbO₂ and oxCCO displayed significant associations between bNIRS channel 14 and PO4. The variability that was observed between chromophores was limited mostly to the non-social condition. For example, only oxCCO displayed significant associations between bNIRS and EEG for the low-frequency theta and α frequency bands. It is well known that various components involved in neurovascular coupling undergo development postnatally, see the review by [18] for a full discussion. Briefly, there is extensive structural change within cerebral microvasculature including growth, extension and proliferation of new blood vessels [19], [20]. Further, studies have also demonstrated gradual development of vascular reactivity (i.e., change in vascular tone, vasoconstriction and vasodilation) [20], [21] which is necessary for the propagation of the NVC response [23]. Lastly, pericytes and astrocytes which are key components of NVC are also known to undergo development in size, number, connectivity and branching [24]–[27]. From the metabolic perspective, infant positron emission tomography (PET) studies demonstrate regional-specific, progressive increase in the cerebral metabolic rate of oxygen consumption (CMRO₂) [14] while others evidence a developmental maturational change in oxidative metabolism [15]. In adults, previous research has also suggested that oxygen consumption is more spatially localised in comparison to changes in cerebral blood flow [10] and that oxCCO has distinct spatial distributions in the brain [11], [12], [13], indicating that energy metabolism may be more spatially specific. The spatial distribution of oxCCO in different brain regions currently remains unmapped in the developing infant brain, however. Therefore, taken together, given that there are extensive changes in cerebral vasculature as well as the metabolic environment and potential variability in the spatial distribution of oxCCO, it is expected that there will be some variability observed in the associations between the haemodynamics and metabolism with neural activity. In our study, we observed more consistent oxCCO – EEG associations across frequency bands and stimuli with more localised (fewer long-range) associations. Further studies with a larger sample size and longitudinal follow up can provide a clearer view on how NVC develops in infancy which will help explain some of the observed variability. Moreover, future studies with high density bNIRS arrays will provide clarification on the spatial distribution of oxCCO in the infant brain.”

Second, about the authors' predictions in the introduction, they mention that they expect to observe "coordinated increases in hemodynamics/metabolism and neural oscillatory activity" in the social condition, and "coordinated decreases in hemodynamics/metabolism and neural oscillatory activity" in the non-social condition. This wording suggests that this is a hypothesis that the authors will be able to test. However, the authors do not test this hypothesis further in the manuscript and it seems to me that the directionality of coordinated activity cannot be directly assessed using this GLM approach. I would therefore suggest to slightly reframe the phrasing of this paragraph to avoid that confusion.

The wording in the introduction and the discussion has been changed to the following and instead of the word coordinated, we have used the word simultaneous.

“We predicted that for the individual comparison of the social condition, we would observe positive associations between bNIRS and EEG measures, i.e. a simultaneous increase in haemodynamics/metabolism and neural oscillatory activity in the β and γ frequency bands (based on previous combined EEG – fMRI studies [1]–[7]) which would be localised to core social brain regions. We hypothesised that for the non-social condition, over the same brain regions, positive associations would be observed between bNIRS and EEG measures, but they would be a simultaneous decrease in haemodynamics/metabolism and oscillatory activity. We also expected simultaneous increases in haemodynamics/metabolism and oscillatory activity to be localised to the parietal brain region.”

Author details

1. Maheen Siddiqui

   Centre for Brain and Cognitive Development, Birkbeck, University of London, London, United Kingdom

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Funding acquisition, Investigation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing

   For correspondence

   m.siddiqui@bbk.ac.uk

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2037-6964

2. Paola Pinti

   Centre for Brain and Cognitive Development, Birkbeck, University of London, London, United Kingdom

   Contribution

   Software, Validation, Visualization, Methodology, Writing – original draft, Writing – review and editing

   Competing interests

   No competing interests declared

3. Sabrina Brigadoi

   1. Department of Development and Social Psychology, University of Padova, Padova, Italy
   2. Department of Information Engineering, University of Padova, Padova, Italy

   Contribution

   Formal analysis, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-3032-7381

4. Sarah Lloyd-Fox

   Department of Psychology, University of Cambridge, Cambridge, United Kingdom

   Contribution

   Conceptualization, Supervision, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

5. Clare E Elwell

   Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom

   Contribution

   Conceptualization, Resources, Software, Supervision, Funding acquisition, Methodology

   Competing interests

   No competing interests declared

6. Mark H Johnson

   Department of Psychology, University of Cambridge, Cambridge, United Kingdom

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Project administration, Writing – review and editing

   Competing interests

   No competing interests declared

7. Ilias Tachtsidis

   Department of Medical Physics and Biomedical Engineering, University College London, London, United Kingdom

   Contribution

   Conceptualization, Resources, Software, Supervision, Funding acquisition, Methodology, Project administration, Writing – review and editing

   Contributed equally with

   Emily JH Jones

   Competing interests

   No competing interests declared

8. Emily JH Jones

   Centre for Brain and Cognitive Development, Birkbeck, University of London, London, United Kingdom

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Validation, Visualization, Methodology, Writing – original draft, Project administration, Writing – review and editing

   Contributed equally with

   Ilias Tachtsidis

   For correspondence

   e.jones@bbk.ac.uk

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-5747-9540

• 555

  Page views

• 113

  Downloads

• 0

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.