TL;DR
By combining two rare datasets of awake infant fMRI, researchers have shown that the human brain’s specialized "face patches" (FFA, STS, and MPFC) are already selective for faces as early as 2 to 5 months of age. This discovery challenges long-held theories that face-processing regions require years of visual experience to develop, suggesting instead that the neural scaffolding for social cognition is remarkably robust shortly after birth.
What Problem Does This Paper Address?
In the adult brain, specific regions are "face-selective," meaning they fire significantly more when we see a face than when we see a house, a hand, or a car. These include the Fusiform Face Area (FFA) for recognition, the Superior Temporal Sulcus (STS) for social cues like gaze, and the Medial Prefrontal Cortex (MPFC) for higher-level social meaning.
A central debate in developmental neuroscience is how these regions get there. One school of thought suggests they are "slow-cooked," requiring years of visual experience and "training" to specialize. Another suggests they emerge very early due to innate biological constraints. Testing this is notoriously difficult: scanning awake infants in an MRI is a logistical nightmare involving high motion, short attention spans, and low signal-to-noise ratios. Consequently, previous studies had small sample sizes that couldn't definitively map the emergence of selectivity across multiple brain lobes in the first year of life.
Methods
The authors tackled the "small sample" problem by aggregating two independent awake infant fMRI datasets, resulting in a sample of 65 infants (aged 2.6–9.6 months)—roughly double the size of previous reports.
Key technical approaches included: * Dynamic Stimuli: Instead of static photos, infants watched 18-second movie blocks of faces, bodies, objects, and scenes to maximize engagement and neural response. * Cross-Validated fROI Analysis: Because infant brain anatomy is highly variable, the researchers didn't rely on a "one-size-fits-all" brain atlas. Instead, they used a functional Region of Interest (fROI) approach: they used half the data to find the "hottest" face-responsive voxels in each individual baby and the other half to independently measure how selective those voxels were. * Motion Scrubbing: They employed rigorous data-cleaning protocols to remove timepoints where the infants moved more than 2mm, a necessity for infant imaging.
Datasets
The study utilized two primary datasets collected at MIT:
- Coil 2011: Data from 31 infants using a custom 32-channel infant head coil designed for a 3T Siemens Trio scanner.
- Coil 2021: Data from 56 infants using a newer, higher-resolution coil and sequence on a 3T Siemens Prisma scanner.
- Final Sample: After exclusions for motion and data quality, 65 infants provided usable whole-brain maps, and 33 unique individuals provided enough data for the more rigorous fROI analyses.
Key Insights and Ideas
- Selectivity vs. Preference: The study didn't just show that babies "like" faces. By comparing faces against other complex categories (like bodies and scenes), they proved true selectivity.
- The OFA Paradox: In adults, the Occipital Face Area (OFA) is thought to be "upstream" of the FFA, processing face parts before the FFA processes the whole face. Surprisingly, this study found that the FFA was selective in the youngest infants (2–5 months), but the OFA was not. This suggests that the "bottom-up" hierarchy we see in adults might not be how the system starts out.
- Social Brain Early Entry: The MPFC—a region associated with complex social evaluations (like "is this person nice?")—showed face selectivity in the youngest group. This suggests that even 3-month-olds might be recruiting "high-level" social machinery when looking at faces.
- Bilateral Beginnings: While adult face processing is usually right-hemisphere dominant, the infants showed robust responses in both hemispheres, suggesting that lateralization is a process that unfolds later in development.
Key Results
- Early Emergence: In infants aged 2–5 months, significant face selectivity (Faces > Bodies, Objects, Scenes) was found in the FFA, STS, and MPFC (all p < 0.05).
- Developmental Lag: The OFA only showed significant face-selective responses in the older group (5–9 months).
- Magnitude: While selectivity was present early, the absolute magnitude of the response was smaller than what is typically seen in adults, indicating that while the "patches" are there, they continue to refine and strengthen throughout childhood.
Limitations and Future Work
- The "First Two Months" Gap: The youngest infants in this sample were ~2 months old. The researchers still cannot say if these regions are selective at birth or if the first 8 weeks of visual experience are the critical "training" period.
- Cross-Sectional vs. Longitudinal: The study compared different babies at different ages. A longitudinal study following the same infants from birth through the first year would provide a clearer picture of individual developmental trajectories.
- Functional Interpretation: While we know the MPFC is active, we don't know what it is doing. Future work could investigate whether these early responses in the MPFC are purely visual or if they already carry the "social-evaluative" weight seen in adults.
Summary updated