When the first social smile actually appears: developmental timing patterns from 24,832 milestone logs

The CDC and AAP both anchor the "first social smile" at the two-month visit. In a cohort of nearly 25,000 Wermom App users, the population median lands earlier — but the normal band is wider than the milestone checkboxes suggest, and the tail extends past the moment many parents start to worry.

1. Why the first social smile is the milestone parents fixate on

Of all the early infant milestones, the first responsive social smile is the one most parents can identify the moment it happens — and the one that produces the most "is something wrong?" questions when it doesn't. There is a developmental reason for both. The first social smile marks the behavioral emergence of what developmental psychologists call dyadic affective engagement: the infant is no longer reflexively responding to internal states but is recognizing, processing, and responding to a familiar caregiver's face and voice⁵. It is the first observable evidence that the social brain is online.

Because of this, the AAP's Bright Futures supervision schedule treats the social smile as an anchor milestone for the 2-month well-child visit³, and the CDC's 2022 update of the "Learn the Signs. Act Early." milestone checklists keeps it in the 2-month list, with an expectation that 75% or more of children should reach it by the checkpoint¹. The 2022 update was significant: it moved the framing from "by 2 months" to a more nuanced "by the 2-month checklist age range," reflecting the evidence that the milestone naturally appears across a window rather than at a single point.

The question for parents and pediatricians is then a quantitative one: across a large contemporary cohort, where exactly does that window sit?

2. The population curve: week 6.4 median, week 9 outer edge

Across the full cohort, the median age at first parent-confirmed responsive social smile was week 6.4, with a 25-to-75 interquartile band running from week 5.1 to week 7.7. The 10th percentile was week 4.3 — meaning roughly 1 in 10 infants in this cohort produced a confirmed social smile before the end of week 4. The 90th percentile was week 9.1, and the 95th percentile was week 10.3.

What this curve tells parents and pediatricians is two things at once. First, the headline framing is fair: by the 2-month well visit, the great majority of healthy term infants have produced a recognizable responsive smile, consistent with the CDC's "75% by this age" framing¹. Second, the lower bound is meaningfully earlier than 2 months — the first quartile is already smiling by week 5, and a non-trivial fraction by week 4. Pediatricians who hear "she smiled at me when she was 3 weeks old!" from a parent at the 1-month visit can take it at face value; it is uncommon, but not implausibly so, and it is unambiguously normal.

The upper tail is the part with the most clinical signal. Roughly 5% of infants in this cohort logged their first responsive smile after week 10. The AAP's developmental surveillance algorithm flags absence of a responsive social smile at the 2-month visit as one of several "surveillance concerns" that warrant follow-up — not immediate referral, but a closer look and earlier rescreening⁶. Our data is consistent with that approach: at week 8, about 78% of infants had logged a first smile; at week 10, about 95% had; at week 12, nearly 99%. The clinical question is not "is week 9 normal?" (it is) but "is week 12 normal?" (rare enough to warrant a closer look).

3. Adjusting for prematurity: corrected age does most of the work

Within the cohort, 3,142 infants were logged at 37–38 weeks gestation. We re-ran the analysis on this subgroup against corrected age — that is, days since the 40-week due date rather than days since birth.

Corrected-age results for the late-preterm subgroup overlapped almost perfectly with the term-only median: corrected median of week 6.6, 25-to-75 band of 5.2 to 7.9. This is consistent with the broader developmental literature, which treats neurobiological milestones in late-preterm infants as a function of post-menstrual age rather than chronological age for at least the first 6 months⁷. Parents of late-preterm infants who are anxious because "everyone else's 8-week-old is smiling" can be reassured: the right comparison is to the post-due-date age, and on that clock the timing is the same.

For early preterm infants (<37 weeks gestation), the sample size in this cohort was insufficient for stable percentile estimates and we did not report subgroup statistics. The AAP and NICHD's guidance for preterm developmental tracking remains the appropriate reference for those families⁷.

4. The "real first smile" vs the "gas smile" — a definitional aside that matters

The single most common reason a parent later "corrects" their first-smile date is the realization, in hindsight, that the smile they logged in week 2 was not socially responsive. Newborn smiles in the first 4 weeks are well-documented in the developmental literature as REM-associated or post-feeding facial movements without dyadic engagement — what Wolff's classic 1968 work called "endogenous smiles"⁴. They look like smiles. They are not the developmental milestone.

Our app's in-product educational prompt asks parents three questions before logging the milestone: (a) Did the smile appear in response to your face or voice? (b) Did it happen while the infant was alert and calm — not feeding, not falling asleep? (c) Has it happened more than once in similar conditions? Only milestones meeting all three confirmations are counted. This is a notable methodological tightening compared to general parent-survey data on early milestones, where definitional ambiguity tends to pull the median earlier than the objective measure justifies.

5. The role of face-time hours: a descriptive pattern, not a causal claim

Within the cohort, parents who logged structured face-to-face engagement (using the Wermom App's "interaction time" feature, available to a subset of users) averaging more than 90 minutes per day during weeks 3–5 had a slightly earlier median first smile (week 6.0) than parents who logged less than 45 minutes per day in the same window (week 6.8). This is a 0.8-week difference at the median and is statistically detectable in this cohort size.

We want to be very careful about how this is interpreted. The 0.8-week difference is real in the data but small compared with the natural variation in the cohort, and it is almost certainly entangled with selection: parents who track interaction time at all are likely to be more developmentally engaged on multiple unmeasured dimensions. Messinger and Fogel's 2007 review of social smile development emphasizes that the smile emerges through a bidirectional interaction process and that the timing is sensitive to the responsiveness of the social environment⁵. Our descriptive data is consistent with that picture without proving causation.

The practical guidance from this is unchanged: the AAP's Bright Futures anticipatory guidance recommendation for high-quality face-to-face interaction with infants from the newborn period onward stands, with or without our data³. Parents reading this cohort study should not interpret it as a prescription. The smile arrives when the developmental machinery is ready, and engaged caregiving across the first weeks is the well-supported context in which that readiness becomes a behavior.

6. When absence of a social smile warrants attention

The AAP Council on Children with Disabilities published a developmental surveillance and screening algorithm that has been reaffirmed through 2024⁶. Within that algorithm, absence of a responsive social smile at the 2-month visit is a documented surveillance concern. It does not automatically trigger referral; it does trigger closer follow-up at the next well visit, with re-evaluation at 4 months. Our cohort data is consistent with this framing.

What our data adds is more granularity on the natural upper tail. By week 10, roughly 95% of healthy term infants in our cohort had produced a confirmed social smile. By week 12, the figure was 99%. The clinical implication is that pediatricians can use the 4-month visit (week 16) as a meaningful re-evaluation checkpoint for the small fraction of infants who had not yet smiled at the 2-month visit. A term infant who is alert, growing, feeding normally, and beginning to show other 2-month social signs (sustained eye contact, cooing, head-orienting to a parent's voice) but whose responsive smile arrives at week 9 or 10 is, by the data, normal.

Persistent absence past 12 weeks in a term infant, particularly when paired with limited eye contact, low overall responsiveness, or feeding/tone concerns, is a different picture and is precisely what the AAP surveillance algorithm is designed to catch⁶. The accompanying parent-facing companion piece on early autism signs at 12–18 months covers the later observational window; the relevant action at 2–4 months is good pediatric surveillance, not parent self-diagnosis.

7. What this data does — and does not — say

This is descriptive epidemiology of a self-selected app-using cohort. It is not a population-representative survey, not a developmental trial, and not a substitute for the standardized milestone instruments (Ages and Stages Questionnaire, Survey of Wellbeing of Young Children, etc.) that pediatricians use in practice. Three caveats deserve attention:

Self-selection. Wermom App users skew toward parents with higher digital engagement, more day-to-day tracking interest, and — based on app analytics — geographically toward urban areas in the US, UK, Canada, and Australia. There is no obvious reason to expect a meaningful population difference in social smile timing specifically (this is a robust neurobiological milestone), but cross-cultural and cross-socioeconomic differences in milestone reporting are well-documented in the broader literature.

Definition sensitivity. The three-question confirmation we used tightens the definition compared to parent-survey data. The trade-off is a slightly later median (by perhaps 0.5–1.0 week) than a permissive single-question survey would report. The advantage is a measurement that aligns better with the developmental construct.

Causal inference. The face-time interaction finding is descriptive and likely confounded. We do not infer that more interaction causes earlier smiles. We do think the broader pattern — engaged, responsive caregiving across weeks 1–6 — is part of the developmental scaffolding for the smile, consistent with the developmental literature⁵.

8. Limitations and future research

Beyond the caveats above, the most important methodological note is that this analysis captures the first confirmed smile, not the stabilization of social smiling as a sustained behavior. Some infants in the cohort logged a first smile, then went a week or two before the next, before settling into daily social smiling. A future analysis will characterize the lag between first confirmed smile and "daily smile" stability and ask whether that lag has clinical signal independent of the first-smile timing.

We also plan to release a companion analysis pairing this milestone with the next two anchor milestones in the AAP Bright Futures schedule — sustained head control (around 4 months) and the social laugh (around 4–5 months) — to characterize whether early or late social-smile timing predicts the trajectory of subsequent milestones. That work is in pre-registration as of May 2026.