Social Smiling and Laughter Are Linked to Enhanced Functional Brain Connectivity in Young Infants’ Default Mode Network

Abstract

The current study used functional near‐infrared spectroscopy (fNIRS) to investigate whether and how individual differences in positive social engagement among 5‐month‐old (N = 109; N = 35 final sample) infants relate to variability in functional connectivity in the human brain's Default‐Mode Network (DMN). Neuroimaging results showed that on average infants displayed greater functional connectivity in the right than in the left hemisphere of the DMN, adding to prior work indicating faster connectivity development in the right hemisphere. Results did not show any positive associations between our preregistered measures of positive social engagement and functional connectivity in the DMN. However, an additional analysis revealed that higher levels of infants’ smiling and laughter during daily social interactions with their caregivers positively predicted DMN functional connectivity in the left hemisphere. This suggests that individual differences in connectivity in a long‐range brain network implicated in a host of social and cognitive functions are associated with some aspects of infants’ positive social‐interactive behaviors.

1. Introduction

Humans are ultra‐social beings (Tomasello 2014), possessing a unique suite of early‐developing social and cognitive skills (Tomasello 2019). Underpinning these early‐developing, social–cognitive skills appears to be a unique motivation to interact with and learn from others (Csibra and Gergely 2009; Tomasello 2020). This social motivation is considered to be a set of processes that help humans function in collaboration with others through orienting to the social world, finding pleasure in social interactions, and maintaining social relationships (Chevallier et al. 2012). On the behavioral level, social engagement is a manifestation of the underlying social motivation of an individual (Chevallier et al. 2012; Over 2016), wherein individuals’ behaviors, such as visual attention, approach, touch, and positive affect, reflect their interest in and motivation to interact with social others. In particular, affective displays are one of the first nonverbal social behaviors used by infants to communicate with others, with social smiles and laughter emerging between 2 and 3 months (Messinger and Fogel 2007; Mireault et al. 2012). Reduced social engagement behaviors, and most notably social smiling, have important developmental implications and have been identified as one of the early emerging indicators of Autism Spectrum Disorder (Nichols et al. 2014; Parlade et al. 2009; Yirmiya et al. 2006). Much of the research to date has focused on the associations between early social engagement in infancy and social, behavioral, and cognitive outcomes in toddlerhood or childhood (Salley et al. 2016; Van Hecke et al. 2007). However, little research has focused on whether and how social behavioral engagement and motivation among young infants are linked to functional brain network development.

The Default Mode Network (DMN) is comprised of various brain areas, including among others, the medial prefrontal cortex (mPFC) and superior temporal cortex (STC), implicated in a host of social and cognitive brain functions (Buckner and DiNicola 2019). The DMN is measured as correlations in spontaneous brain fluctuations typically seen during rest or when the brain is focused on internal thoughts and reflections during mind wandering (Raichle 2015). Research with adults suggests this network plays a key role in maintaining a continuous and stable sense of self (Buckner et al. 2008; Northoff et al. 2006). Moreover, the DMN is comprised of key social brain regions (Mars et al. 2012; Schilbach et al. 2008), suggesting this network has shared involvement in self‐referential and social information processing (Grossmann 2025). Research using resting‐state functional magnetic resonance imaging (fMRI) and functional near‐infrared spectroscopy (fNIRS) shows that functional connectivity within the DMN, although less specialized than in adults, is already present in newborn infants and may be one of the first higher‐order functional networks to develop during infancy (Gao et al. 2009; Hu et al. 2022; Kelsey, Farris, et al. 2021; Yu et al. 2023). Importantly, individual brain areas thought to be part of the DMN, such as the mPFC and STC, have been implicated in a series of social processes. These processes include responding to self‐relevant social interactive cues such as eye contact, smiles, own name, or infant‐directed speech, developing around 4–5 months (see Grossmann [2015, 2017]). For example, viewing dynamically changing faces with shifting eye gaze toward the infant followed by a smile has been associated with enhanced responses in the medial prefrontal and STC (see Grossmann et al. 2008). Enhanced activation of these two brain regions during self‐relevant social information processing in infancy suggests functional specialization within brain areas also implicated in the DMN.

The mPFC, involved in mentalizing or theory‐of‐mind in adults (Amodio and Frith 2006), has been shown in infants to selectively respond to viewing smiles compared to frowns, hearing infant‐directed speech compared to adult‐directed speech, and both hearing and viewing one's own mother compared to another female (Grossmann 2013; Krol and Grossmann 2020; Naoi et al. 2012; Saito et al. 2007). The mPFC may reflect infants’ interpretations of smiles as a positive communicative signal (Grossmann et al. 2008), supported by research demonstrating that the mPFC preferentially responds to positively valenced social stimuli early in infancy, which continues into toddlerhood (Grossmann 2013; Powell et al. 2018; Richardson et al. 2021). Importantly, recent research suggests that mPFC responses to social smiles at 11 months, but not social frowns, longitudinally predict social motivation behaviors at 18 months (Grossmann and Allison 2024). Overall, converging evidence suggests the mPFC plays an important role in developing social brain function early in ontogeny and may promote preferences for positive social engagement.

It has also been shown that newborn infants engage superior temporal regions when viewing dynamic faces (Farroni et al. 2013), and by 5 months, bilaterally engage posterior superior temporal regions in response to social video clips (Lloyd‐Fox et al. 2014). Importantly, responses of superior temporal regions while viewing dynamic social stimuli increase with age (and social experience), indicating the role of face‐to‐face engagement in developing the social function of this bilateral region (Farroni et al. 2013). Recent research suggests that enhanced right STC responses to changing faces at 7 months longitudinally predict social behavior (seeking out and enjoying social interactions) at 18 months (Grossmann 2024). Together, this suggests that the STC is involved in infants’ processing of social information and may play a role in promoting positive social engagement.

Taken together, existing neuroimaging research suggests that the mPFC and the STC (1) develop social functions within the first few months of postnatal life, (2) are positively associated with increased positive social interaction, and (3) are longitudinally associated with overt social behavior in the second year. However, it is currently unclear whether and how the coordination of functional connectivity between these brain regions that are part of the DMN is linked to variability in infants’ positive social engagement before 7 months.

Functional brain connectivity is considered to be one of the most promising avenues for understanding brain–behavior relationships (Friston 2011). Although fNIRS offers a limited approximation of the full DMN, it has been used to observe long‐range functional connectivity in various resting‐state networks, including cortical components of the DMN, in young infants and has been linked to infants’ affect and behavioral measures (Z. Fu et al. 2023; Kelsey, Farris, et al. 2021). In adults, research suggests changes in DMN connectivity are associated with changes in social behavior (Breukelaar et al. 2020; Li et al. 2014; Mars et al. 2012). Previous work shows that, at 5 months, increased maternal sensitivity measured during a mother–infant free play interaction was associated with enhanced functional connectivity in the DMN (Chajes et al. 2022), suggesting that maternal positive engagement has an effect on the DMN in the developing infant.

Extending this line of research, the current study tested our preregistered hypothesis (https://osf.io/3bs4v) that heightened levels of infant positive social engagement (social attentiveness, social engagement, social smiling and laughter, social touch, and social approach) are positively associated with resting‐state functional connectivity in DMN among a sample of 5‐month‐old infants.

2. Materials and Methods

2.1. Participants

Participating infants were part of a larger longitudinal study of social and emotional development. Mother–infant dyads (N = 121) were first recruited from a local hospital when the infants were newborns (for details, see Kelsey, Farris, et al. [2021] and Kelsey, Prescott, et al. [2021]). To be included, participants had to be born at term, with normal birth weight (> 2500 g), and not have any hearing or visual impairments. Of the original 121 dyads, 109 families returned to the lab when the infants were 5 months old (M age = 5.2 months; SD = 0.68 months; range = 4–7 months; n = 64 male, sex assigned at birth). At the 5‐month‐old timepoint, infants completed a 5‐min free play session with their mothers following a resting‐state fNIRS recording session. Additionally, mothers completed questionnaires that included questions about their infants’ behavior. Of the 109 dyads that completed the 5‐month visit, 35 were included in the final analytic sample (M age = 5.8 months; SD = 0.44 months; range = 4.9–6.8 months; n = 24 male sex assigned at birth) who had all variables of interest detailed below. Seventy‐four participants were excluded from the present analyses for one or more of the following reasons: n = 45 excluded for having more than 50% of fNIRS channels excluded during preprocessing (see below for details); n = 16 excluded due to technical errors; n = 7 excluded because they did not have at least 100 s of continuous fNIRS data with no disruptive behaviors (see below); n = 1 excluded because of inaccurate placement of the fNIRS cap (more than 1 cm deviation from proper cap placement); n = 21 excluded for not completing item 90 of the Infant Behavior Questionnaire‐Revised (IBQ‐R) short form or answering “N/A” or “prefer not to answer”; n = 2 excluded for not completing free play interaction; n = 1 excluded for deep sleep or crying for more than 50% of the free play interaction. Neuroimaging‐related exclusion rates (63%) were higher compared to prior infant fNIRS studies (Baek et al. 2023). However, representation of sex did not significantly differ based on overall exclusion (χ ² [1, N = 109] = 1.51, p = 0.219). Additionally, we tested whether infants who had parent‐reported (PR) data but did not contribute fNIRS data statistically differed from infants who contributed both PR and fNIRS data. Welch two‐sample t‐tests revealed there was no significant difference in PR‐Smiling and Laughing scores (see variable descriptions below) between infants who had PR data but did not contribute fNIRS data (M = 3.93, SD = 1.17) and those who contributed both fNIRS and PR data (M = 4.09, SD = 1.31), t[74.65] = −0.61, p = 0.544. Similarly, there was no significant difference in PR‐Social Approach scores (see variable descriptions below) between infants who only contributed PR data (M = 5.32, SD = 1.67) and those who contributed both fNIRS and PR data (M = 4.95, SD = 1.49), t[83.80] = 1.10, p = 0.273 (see Supporting Information). Parents provided informed consent on behalf of themselves and their infant; all procedures were approved by the authors’ institutional review board, and participants received monetary compensation for their participation.

2.2. Behavioral Coding From Free Play

Infants’ lab‐coded (LC)‐Social Attentiveness, Social Engagement, Social Smiles and Laughter, and Social Touch were assessed using video recordings of mother–infant interactions during the 5‐month timepoint. In order to capture a natural play session, each mother was instructed to interact with her infant as she did at home. Two cameras simultaneously recorded the interaction. One camera captured the face and body of the mother, and the other camera captured the face and body of the infant. Mothers were made aware of the cameras and instructed to remain within view of them. Infants were placed on their backs on a carpet in the center of the playroom, and no further instructions were given. The same selection of four objects (three toys and one playbook) was provided to each mother–infant dyad to assist play. Mothers and infants could freely choose and change the object of interest during the interaction or abandon the objects altogether. Once assembled, the experimenter left the room, and the mother–infant pair was left alone to interact for 5 min. LC‐Social Attentiveness, LC‐Social Engagement, LC‐Smiling and Laughing, and LC‐Social Touch were behaviorally coded and scored offline using INTERACT software (Mangold International, Arnstorf, Germany), and the exact coding scheme can be accessed through OSF (https://osf.io/preprints/psyarxiv/ym3ne) (see Grossmann et al. [2018] and Krol et al. [2019] for information on the initial creation of this coding scheme). LC‐Social Attentiveness and LC‐Social Engagement were scored on a scale from 1 (not at all) to 5 (very much). LC‐Social Attentiveness and LC‐Social Engagement were coded separately to acquire more detailed information about infants' behaviors: (1) How attentive was the infant toward the mother? and (2) How engaged was the infant toward the mother? Since scores were strongly correlated (Pearson's r = 0.57), LC‐Social Attentiveness and LC‐Social Engagement scores were combined into one LC‐Social Engagement score, consistent with previous work by Grossmann et al. (2018). LC‐Social Attentiveness and LC‐Social Engagement were double‐coded for all videos by two independent raters for reliability. Inter‐rater reliability was good for LC‐Social Attentiveness (Intraclass correlation coefficient = 0.950; Fleiss Kappa = 0.774) and LC‐Social Engagement (Intraclass correlation coefficient = 0.922; Fleiss Kappa = 0.694). Discrepancies were resolved via conferencing. LC‐Social Touch and LC‐Smiling and Laughing were scored based on the duration of time infants exhibited the behavior during the 5‐min free play interaction. Specifically, LC‐Social Touch only included infant‐initiated touch directed at the mother (i.e., proactively touching their mother), and LC‐Smiling and Laughing only included infants’ smiles and laughter directed at the mother (i.e., smiling or laughing while looking at their mother or as a direct response to their mother). Infant LC‐Smiling and Laughing duration were coded together since infant smiles typically occurred with laughter.

2.3. Parental Report of Infant Behavior

PR‐Social Approach was measured using the Approach subscale of the PR IBQ‐R short form (Putnam et al. 2014). Mothers completed a series of questionnaires on Qualtrics (Provo, UT) before each timepoint of data collection. The IBQ‐R short form was completed by mothers before in‐person data collection at the 5‐month‐old timepoint. Items were rated by mothers on a scale from 1 to 7 (1 [never]–7 [always]) with two additional answer options provided (8 [N/A] and 9 [prefer not to answer]). The following item on the Approach subscale (item 90) was selected: “When familiar relatives/friends visited, how often did the baby get excited?” Although only one item on the Approach subscale represented our PR‐Social Approach variable, this item is similar to the Sociability scale items from the Early Childhood Behavior Questionnaire administered between 18 and 36 months of age (Putnam et al. 2006). The Sociability subscale has been used in previous work to capture social motivation behaviors (i.e., seeking out and enjoying social interactions with others; see Grossmann and Allison [2024]). As an exploratory question, we separately included a PR‐Smiling and Laughing variable using the Smiling and Laughter subscale (6‐item composite score) from the IBQ‐R short form (Cronbach's alpha value for 3–6 months of age = 0.85), defined as smiling and laughter from the child in general caretaking and play situations (see Table 1), to increase power of PR positive social engagement (Gartstein and Rothbart 2003; Putnam et al. 2014). Additionally, previous research shows that smiling and laughter assessed in a laboratory is not associated with PR assessments of smiling and laughter within the first 6 months (Planalp et al. 2017).

TABLE 1.

This table lists the six items taken from the IBQ‐R short form used to determine the Smiling and Laughing subscale score in the current study.

Item #	Infant Behavior Questionnaire‐Revised short form Smiling and Laughing subscale items
40	When being dressed or undressed during the last week, how often did the baby smile or laugh?
42	When put into the bath water, how often did the baby smile?
43	When put into the bath water, how often did the baby laugh?
65	When face was washed, how often did the baby smile or laugh?
11	How often during the last week did the baby smile or laugh after accomplishing something (e.g., stacking blocks, etc.)?
12	How often during the last week did the baby smile or laugh when given a toy?

2.4. fNIRS Data Recording

Infants sat on their parent's lap approximately 60 cm from the 23‐inch monitor that displayed the resting‐state video stimulus for a total of up to 7 min while fNIRS data were being recorded. The infants wore an fNIRS fabric cap (EasyCap, Germany), which was secured in place using infant overalls and outside netting. Parents were asked to remain quiet throughout the fNIRS recording session. Sessions were video‐recorded, and a trained research assistant behaviorally coded infants’ behavior during the fNIRS recording offline using the behavior coding scheme from Kelsey, Farris, et al. (2021).

fNIRS data were recorded using an NIRx NIRScout system and NIRStar acquisition software. The fNIRS system has 49 channels (approximately 2 cm source‐detector distance) covering the frontal, temporal, and parietal brain regions in both hemispheres (see Chajes et al. [2022], Grossmann et al. [2018], and Krol et al. [2019]). The system emits two wavelengths of light, 760 and 850 nm, and captures both oxygenated hemoglobin (oxyHb) and deoxygenated hemoglobin (deoxyHb). The diodes have a power of 25 mW/wavelength, and data were recorded at a preset default sampling rate of 3.91 Hz.

2.5. fNIRS Video Stimuli

Following previously published work (Kelsey, Farris, et al. 2021; Kelsey, Prescott, et al. 2021), the nonsocial stimulus was created by selecting nonsocial video clips (e.g., toys, fruits, and everyday objects) from a popular infant video (Baby Einstein—Kids2 Inc.). The images were accompanied by classical music, and the video was segmented into 30‐s intervals. The order of video segment presentation was randomized for each infant.

2.6. fNIRS Data Processing

Timepoints were removed if infants were crying, looking at the parent or experimenter, or parents were talking. A primary coder coded all of the videos, and an additional trained research assistant coded a selected subsample of videos for reliability (28.4%, n = 31). Interrater reliability was excellent (Intraclass correlation coefficient = 0.92). Based on previous work, infants needed a minimum of 100 s of disruption‐free data to be included (see Chajes et al. [2022]). Following our preregistered fNIRS data processing plan, custom Matlab scripts and Homer2 were used to preprocess the data in accordance with guidelines outlined in Powell (2020). First, raw intensity data were converted to optical density units, and channels with mean intensities outside the system‐recommended values (enPrunechannels: dmin = 10–2, dmax = 109) or high‐frequency amplitude changes with over 90% of timepoints marked as motion (hmrMotionArtifactByChannel_indLambda: tMotion = 1.0, tMask = 1.0, stdThresh = 100, ampThresh = 0.1) were removed. Flexible targeted principal component analysis with up to three iterations (tMotion = 1.0, tMask = 1.0, Std Thresh = 100, Amp Thresh = 0.2, tpcaFilter = 0.97) was used to correct for motion artifacts. Consistent with previous infant fNIRS studies, corrected data were band‐pass filtered from 0.01 to 0.08 Hz. This frequency range is often selected as it encompasses frequencies thought to correlate with neuronal activity (0.025 Hz) while minimizing contributions from unwanted signal components such as Mayer waves, breathing rate, and heart rate that occur at approximately 0.09, 0.25, and 1.3 Hz, respectively (Bulgarelli et al. 2020; Chajes et al. 2022; Kelsey, Farris, et al. 2021; Pinti et al. 2019). Data were converted into oxyHb and deoxyHb concentration change values using a modified Beer–Lambert Law, assuming a partial path length factor of five (hmrOD2Conc).

2.7. Computing Functional Network Scores

For each infant, a 49 × 49 correlation matrix was created depicting all of the Fisher Z‐transformed correlation values between all of the channels measured. The DMN was created from an average of select channels that corresponded with regions of interest (Chajes et al. 2022; Kelsey, Farris, et al. 2021; Kelsey, Prescott, et al. 2021). Brain areas were named in accordance with anatomical mappings of the 10–20 system in similar‐age infants. Based on the LONI probabilistic brain atlas (LPBA, Shattuck et al. 2008) using photon propagation simulation with realistic, age‐appropriate (6 months) head models (devfOLD; X. Fu and Richards 2021; Zimeo Morais et al. 2018), brain areas were confirmed at the group level. The DMN left and right oxyHb z‐scores were created by averaging the left and right hemisphere correlations between three channels in the mPFC, corresponding with the Fpz electrode (10–20 system) or the superior and middle frontal gyri (LPBA), and three channels in the left and right lateral temporal cortex, corresponding with the T7 and T8 electrodes (10–20 system) or the superior and middle temporal gyri (LPBA). Specifically, left DMN resting‐state functional connectivity network z‐scores were created using the three medial prefrontal and three left hemisphere temporal source‐detector pairs. Right DMN resting‐state functional connectivity network z‐scores were created using the three medial prefrontal and three right hemisphere temporal source‐detector pairs (see Figure 1).

FIGURE 1.

This shows the fNIRS channels used to create the Default Mode Network with reference to the 10–10 system (light blue dots and corresponding labels). Each channel is represented as a circle with white labels representing source‐detector information (nose pointing up). The black lines represent all the correlation values computed between medial prefrontal cortex regions (dark blue) and left and right lateral temporal cortex regions (orange) to create the averaged left and right DMN functional connectivity scores.

2.8. Analysis Plan

Following our preregistered data analysis plan (https://osf.io/3bs4v), multiple linear regression modeling using the entry method was employed to test our main hypothesis that enhanced positive social engagement (LC‐Social Attentiveness + LC‐Social Engagement, LC‐Social Smiling and Laughter, LC‐Social Touch, and PR‐Social Approach) is positively associated with functional connectivity in the DMN. Analyses were carried out in IBM SPSS Statistics (Version 28), and assumptions were tested using R Studio (Version 2024.04). Unlike p‐values, Bayes factors offer a direct measure of the strength of evidence for (BF₁₀ > 1) and against (BF₁₀ < 1) a specific hypothesis, and are especially useful in small sample studies (Coon et al. 2025; Kelter 2020). Therefore, we also employed a complementary Bayesian regression analysis in JASP (Version 0.18.3) to test the extent to which the data showed support for our hypotheses, helping with the interpretation of the statistical findings.

Before conducting the multiple linear models, we first examined whether DMN connectivity differed between the left and right hemispheres using a paired‐samples t‐test. Based on the results, we employed two models to test whether positive social engagement predictor variables were positively associated with functional connectivity of the DMN, one for the left and another for the right hemisphere. Second, the linearity, normality, homoscedasticity, and absence of multicollinearity assumptions of both left and right hemisphere models were checked. Residuals were examined through diagnostic plots. The right hemisphere DMN showed a negative autocorrelation in the residuals (autocorrelation = −0.44, D–W statistic = 2.86, p = 0.002). All other assumption testing indicated that the two models accurately captured the relation between the predictors and the outcome variables, and estimated coefficients for predictors were stable and reliable. Third, as an exploratory analysis, a series of Spearman's rho correlations was used to identify significant associations between the main variables of interest, data quality measures, and demographics. Any significantly identified covariates were included in the models.

3. Results

3.1. Lateralization of DMN

Results of the paired‐samples t‐test revealed a significant difference between left (N = 49 observations) and right (N = 54 observations) DMN connectivity strength (t[47] = −2.09, p = 0.042), with the right hemisphere DMN (M = 0.28, SD = 0.28) showing significantly greater functional connectivity than the left hemisphere (M = 0.18, SD = 0.20) (see Figure 2). Results of one‐sample Student t‐tests revealed that both left (t [48] = 6.06, p < 0.001) and right (t [53] = 7.28, p < 0.001) DMN connectivity values were significantly different from zero, showing that infants in our sample on average demonstrated significant connectivity in this network in both hemispheres (see Supporting Information). We therefore tested whether our positive social engagement predictor variables were associated with left and right DMN functional connectivity in two separate multiple linear regression models.

FIGURE 2.

This raincloud plot shows the difference in functional connectivity in the DMN between the left (green) and the right (orange) hemispheres.

3.2. Univariate Associations Between Data Quality Measures, Variables of Interest, and Demographics

Correlations revealed that the LC‐Social Engagement combined score (Social Attentiveness + Social Engagement) was positively associated with LC‐Smiling and Laughing duration (r [33] = 0.47, p = 0.004) and LC‐Social Touch (r [33] = 0.28, p = 0.049). Interestingly, our PR‐Social Approach score was not significantly correlated with any of the lab‐based positive social engagement variables behaviorally coded from the free play (all ps > 0.05). Additionally, PR‐Smiling and Laughing scores were not significantly correlated with LC‐Smiling and Laughing duration (r[33] = 0.11, p = 0.494), or any other LC positive social engagement behaviors during the mother–infant free play interaction. Interestingly, PR‐Social Approach and PR‐Smiling and Laughing were significantly correlated (r [33] = 0.32, p = 0.031). See Table 2.

TABLE 2.

This table lists the Pearson correlational values between all study variables, their means (M), standard deviations (SD), and respective p‐values.

Study variable correlations (N = 35)
		LDMN	RDMN	LC‐Social Engagement Combined	LC‐Smiling and Laughing	LC‐Social Touch	PR‐Social Approach	PR‐Smiling and Laughing
LDMN (M = 0.18; SD = 0.18)	Pearson	1	0.233	0.090	0.081	−0.077	−0.095	0.459**
	p‐value		0.089	0.303	0.321	0.330	0.294	0.003
RDMN (M = 0.29; SD = 0.22)	Pearson	0.233	1	−0.017	−0.206	−0.014	−0.037	0.006
	p‐value	0.089		0.461	0.118	0.467	0.416	0.486
LC‐Social Engagement Combined (M = 5.26; SD = 1.69)	Pearson	0.090	−0.017	1	0.471**	0.284*	−0.039	0.086
	p‐value	0.303	0.461		0.002	0.049	0.411	0.312
LC‐Smiling and Laughing (M = 8.27; SD = 12.00)	Pearson	0.081	−0.206	0.471**	1	0.238	0.109	−0.004
	p‐value	0.321	0.118	0.002		0.084	0.267	0.491
LC‐Social Touch (M = 27.63; SD = 27.02)	Pearson	−0.077	−0.014	0.284*	0.238	1	0.182	0.060
	p‐value	0.330	0.467	0.049	0.084		0.147	0.367
PR‐Social Approach (M = 5.11; SD = 1.35)	Pearson	−0.095	−0.037	−0.039	0.109	0.182	1	0.319*
	p‐value	0.294	0.416	0.411	0.267	0.147		0.031
PR‐Smiling and Laughing (M = 4.19; SD = 1.29)	Pearson	0.459**	0.006	0.086	−0.004	0.060	0.319*	1
	p‐value	0.003	0.486	0.312	0.491	0.367	0.031

**Correlation is significant at the 0.01 level.

*Correlation is significant at the 0.05 level.

We also tested whether any of the demographic and fNIRS data quality measures, including signal‐to‐noise ratio for wavelength 1/760 nm (SNR 1; M = 10.70, SD = 6.30), signal‐to‐noise ratio for wavelength 2/850 nm (SNR 2; M = 9.84, SD = 7.51), included channels (M = 38.53, SD = 6.77), and included time (M = 343.61, SD = 96.33), were associated with left and right DMN functional connectivity. Of these measures, only SNR 1 (r [42] = 0.35, p = 0.015) and SNR 2 (r [42] = 0.43, p = 0.002) were significantly positively associated with left DMN connectivity. SNR 1 and SNR 2 were strongly positively correlated (r [43] = 0.80, p < 0.001). SNR quantifies the signal quality of fNIRS channels, with lower SNR potentially confounding neural signal estimation. Variations in signal quality are considered to be associated with the computed concentration values (Truong et al. 2022). To account for this potential confound, in addition to the preregistered analyses, which did not include SNR, we ran further analyses including SNR 2 as a covariate. SNR 2 for wavelength 850 nm was chosen as the covariate because the higher wavelength (lower frequency) best reflects signal changes related to oxyHb, which was the fNIRS measure used to compute DMN functional connectivity.

3.3. Preregistered Analysis

Results of the multiple linear models revealed that none of the predictor variables (LC‐Social Attentiveness + LC‐Social Engagement, LC‐Social Smiling and Laughter, LC‐Social Touch, and PR‐Social Approach) significantly predicted DMN functional connectivity in the left hemisphere, F(4, 30) = 0.23, p = 0.922, explaining only 3% of the variance (R ² = 0.03, Adjusted R ² = −0.10), or in the right hemisphere, F(4, 30) = 0.40, p = 0.805 (R ² = 0.05, Adjusted R ² = −0.08). Similar results were found in the model controlling for SNR 2. Corresponding to the results using the linear regression approach reported above, the Bayesian analysis showed greater support for the null hypothesis than our alternative hypothesis for all predictor variables in both models, with and without including SNR 2 as a covariate (see Supporting Information).

3.4. Exploratory Frequentist Analysis

As an exploratory analysis, we conducted multiple linear regressions corresponding to the ones reported above, but with the IBQ‐R‐derived Smiling and Laughing scores as an additional predictor variable to test whether PR‐Smiling and Laughing during daily interactions predicted left and right hemisphere DMN functional connectivity. Results of the exploratory analyses revealed a significant positive relation between PR‐Smiling and Laughing scores and left DMN functional connectivity (F(5, 29) = 3.16, p = 0.021) (R ² = 0.35, Adjusted R ² = 0.24, see Figure 3), but not right DMN oxyHb (F(5, 29) = 0.78, p = 0.571) (R ² = 0.12, Adjusted R ² = −0.03) when controlling for SNR 2 as a covariate. Without controlling for SNR 2, there is a marginally significant positive association between PR‐Smiling and Laughing scores and left DMN functional connectivity (F(5, 29) = 2.45, p = 0.057) (R ² = 0.30, Adjusted R ² = 0.18), but not right DMN functional connectivity (F(5, 29) = 0.31, p = 0.902) (R ² = 0.23, Adjusted R ² = −0.11).

FIGURE 3.

This scatter plot shows the results of the regression modeling, revealing a positive association between the residuals of the parent‐reported Smiling and Laughing scores using the IBQ and the residuals of Default Mode Network functional connectivity in the left hemisphere, controlling for the signal‐to‐noise ratio.

3.5. Exploratory Bayesian Analysis

To test the extent of the association between PR‐Smiling and Laughing scores and left DMN functional connectivity, we employed a Bayesian regression analysis in JASP (Version 0.18.3; Coon et al. 2025; Kelter 2020). In line with the exploratory frequentist analysis results, the Bayesian regression analysis revealed a BF₁₀ = 1.00 for Smiling and Laughing scores without SNR 2 as a covariate and BF₁₀ = 2.02 for Smiling and Laughing scores with SNR 2 as a covariate, showing greater support for our hypothesis than the null hypothesis. Together, the PR measures, PR‐Smiling and Laughing scores, and PR‐Social Approach item scores revealed the greatest support for our alternative hypothesis compared to the null hypothesis for functional connectivity of the DMN in the left hemisphere as the outcome variable (BF₁₀ = 3.91 with SNR2 covariate) (see Supporting Information). In line with the preregistered frequentist analysis results, none of the predictors revealed greater support for the alternative hypothesis than the null hypothesis for functional connectivity of the DMN in the right hemisphere as the outcome variable.

4. Discussion

The current study examined the association between positive social engagement behaviors and resting‐state DMN functional connectivity in 5‐month‐old infants. Individual differences in positive social engagement were measured using parental report (PR‐Social Approach and PR‐Smiling and Laughter) as well as through behavioral observation from mother–infant free play (LC‐Social Attentiveness, LC‐Social Engagement, LC‐Social Smiling and Laughter, and LC‐Social Touch). Contrary to our preregistered hypothesis, our results did not show a positive association between behaviorally coded positive social engagement and resting‐state functional connectivity in the DMN. It is important to note that our small sample size, combined with a conservative preregistered modeling approach, may have limited our ability to detect small effects. However, results from an exploratory analysis show that higher PR levels of infants’ smiling and laughter during daily interactions with their caregivers are positively associated with functional connectivity in the DMN within the left hemisphere. This suggests that, already by 5 months of age, infants’ functional connectivity in a long‐range brain network implicated in a host of social and cognitive functions in adults (Breukelaar et al. 2020; Li et al. 2014; Mars et al. 2012) is associated with variability in infants’ positive affect displayed during social interactions in close relationships. In addition to PR‐Smiling and Laughing, results of a Bayesian regression analysis indicate that infants displaying both higher levels of PR‐Social Approach and smiling/laughter show greater functional connectivity of the DMN in the left hemisphere. Together, this pattern of findings provides evidence for the notion that some aspects of infants’ positive affective engagement and social approach behavior are linked to functional connectivity in the DMN.

Although the preregistered hypotheses were not supported, the results of the exploratory analyses still advance our understanding of the relation between positive social engagement and functional brain connectivity early in human development. The current exploratory findings, indexing a positive association between infants’ positive affective engagement in social interactions, are consistent with prior work demonstrating the early developmental emergence of brain systems involved in promoting the formation of positive social relationships (Grossmann 2024; Grossmann and Allison 2024; Grossmann and Wood 2023). Moreover, the current results extend previous findings of a link between enhanced DMN connectivity and behaviors that facilitate positive social relationships across the lifespan (Li et al. 2014; Yeshurun et al. 2021) into early infancy. In line with our findings, it has also been shown in hyperscanning studies that mPFC (part of DMN) coupling between 9‐ and 15‐month‐old infants and their caregivers is associated with positive social engagement behaviors, including smiling during social interactions with caregivers (see Piazza et al. [2020]). Furthermore, our findings may also relate to previous research with adults proposing a so‐called extended social‐affective DMN (Amft et al. 2015). Some regions in the adult DMN, including the mPFC, have been shown to have differential functional connectivity patterns in different emotional contexts during a task‐based functional connectivity paradigm (Sreenivas et al. 2012). Interestingly, brain regions involved in adults’ extended social‐affective DMN at rest such as the medial prefrontal and the lateral temporal cortex are consistently recruited during social‐affective engagement measured during experimental tasks in infants and toddlers, with increased mPFC engagement when viewing smiling faces compared to neutral faces (Grossmann 2013; Powell et al. 2018; Richardson et al. 2021). Considering our findings show that functional connectivity at rest, precisely between these medial prefrontal and lateral temporal brain regions, is predicted by infants’ positive affect displayed during social interactions, it raises the possibility that this association reflects early developing connectivity in the extended social affective DMN. Thus, future research would benefit from comparing task‐related functional connectivity during positive affective (direct gaze, smiling faces), negative affective (direct gaze, angry faces), and neutral social interactions separately to examine this possibility and to better understand the exact nature of the observed association.

The finding that our effect was lateralized to the left hemisphere agrees with research with infants and adults implicating the left hemisphere in positive affect and approach behaviors (see Davidson and Fox [1982]). Furthermore, with respect to the lateralization of the effect to the left hemisphere, it is important to note that our results showed that, at the group level, 5‐month‐old infants displayed greater functional connectivity in the right than in the left hemisphere of the DMN. This group‐level effect is in line with prior work showing differential cortical maturation of the two hemispheres during infancy, with the right hemisphere maturing earlier than the left hemisphere during early ontogeny (Thatcher et al. 1987). In the context of the current findings, this may suggest that infants who display greater positive affect and positive engagement develop functional connectivity earlier (faster maturation) than infants showing reduced positive affect and positive engagement. This tentative suggestion clearly requires further investigation, especially by testing infants at older ages using a similar approach as used in the current study.

The observed association effect is limited to specific behaviors observed by the parents during daily social interactions and not seen during behavioral observation in the laboratory, which is consistent with previous literature showing that PR and laboratory‐based assessments of smiling and laughter are unrelated within the first 6 months (Planalp et al. 2017). Given that our findings using PR social engagement measures align with previous literature, this suggests a limited ability of laboratory‐based measures to tap into young infants’ behavioral variability relevant to the development of this brain network. Therefore, future research would benefit from investigating whether PR or other non‐lab‐based measures of Social Attentiveness, Social Engagement, and Social Touch are associated with DMN functional connectivity within the first 6 months of life. Additionally, the current study was conducted with a small sample size drawn from a Western context, consistent with common limitations in infant neuroimaging research (Ilyka et al. 2021). Specifically, the current study had a high exclusion rate (68%), with 93% of exclusions due to insufficient fNIRS data quality or inability to collect fNIRS data. As outlined in the methods, there were no differences in PR measures between infants that contributed fNIRS data and infants that were not included in the final analysis due to fNIRS data exclusion. Future infant fNIRS study protocols may benefit from using more engaging resting‐state stimuli and from including other resting‐state measures, such as cytochrome‐C‐oxidase (oxCCO) (Gervain et al. 2023). Importantly, while the use of nonsocial visual stimuli is standard and necessary to reduce motion and maintain engagement in infants, it differs from traditional resting‐state paradigms commonly used in adult resting‐state research and may influence the resting‐state signal (Camacho et al. 2020). To increase sample size and representation of diverse sociodemographic, cultural, and racial‐ethnic groups, future work would benefit from leveraging the portability of fNIRS systems. Additionally, fNIRS is limited to measuring surface cortical structures of the brain and, therefore, offers a limited approximation of medial prefrontal and temporal brain regions and cannot effectively measure subcortical regions of the full DMN. Specifically, our measure of the DMN is restricted to surface cortical regions accessible via fNIRS (i.e., lateral temporal and superficial medial frontal areas). Also, asleep/sedated infant fMRI research shows that the posterior cingulate cortex and inferior posterior lobule are subcortical regions that are part of the DMN throughout its early emergence and rapid development in the first 2 years of life (Yu et al. 2023). Structural (inclusion vs. exclusion of subcortical regions) and functional (asleep/sedated vs. awake state) differences between fNIRS and fMRI measurements of DMN functional connectivity may contribute to inconsistent findings between fMRI and fNIRS studies (Agyeman et al. 2023). Finally, the single‐timepoint, correlational design of the current study cannot provide insights about the developmental trajectory of the observed brain–behavior association. Future longitudinal work examining the associations between infants’ daily positive social affective behavior with caregivers and DMN functional connectivity is needed to further our understanding of the nature of the observed association.

5. Conclusion

In summary, the current study provides preliminary evidence that, already by 5 months of age, infants’ functional connectivity in a major long‐range brain network implicated in a host of socio‐affective and cognitive functions is associated with variability in infants’ positive affect displayed during social interactions in close relationships. Moreover, the observed effect is restricted to the left hemisphere, which in agreement with prior research appears to be the hemisphere developing more slowly during infancy and being involved in positive affect and approach. The current findings support the general notion that brain development is intricately connected to positive experiences in close social relationships very early in life.

Author Contributions

O.A. conceptualized the study, analyzed data, and wrote the manuscript. T.G. conceptualized the study, supervised data collection, and provided edits on the manuscript. C.K. completed data collection, analyzed data, and provided edits on the manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary materials for this paper have been posted to OSF and are publicly available at this link: https://osf.io/3bs4v/files/osfstorage.

Allison, O. , Kelsey C., and Grossmann T.. 2025. “Social Smiling and Laughter Are Linked to Enhanced Functional Brain Connectivity in Young Infants’ Default Mode Network.” Developmental Psychobiology 67, no. 5: e70088. 10.1002/dev.70088

Data Availability Statement

Data that support the findings of this study have been posted on OSF and are publicly available at this link: https://osf.io/3bs4v/files/osfstorage.