Characterizing age-related changes in spontaneous fetal brain activity using fMRI
Year: 2026
Type: preprint
Abstract: Brain maturation progresses rapidly across gestation. Billions of neurons develop within the first twelve weeks of gestation (Linderkamp, 2009) and by 20 weeks, they are organized into complex functional networks (Thomason, 2020; Turk et al., 2019). This dramatic neural growth renders the fetal brain highly vulnerable to environmental inputs. Indeed, replicated findings show that intrauterine exposures prospectively correlate with (in humans) and cause (in animals) enduring alterations to neural structure and function (Monk et al., 2019; Van den Bergh et al., 2020). According to the fetal programming hypothesis (Seckl & Holmes, 2007) and the Developmental Origins of Health and Disease framework (DOHaD; Barker, 2004), these prenatally-programmed changes to brain development may in turn confer risk for psychopathology and other neurodevelopmental difficulties across the lifespan (Di Martino et al., 2014). Thus, gestation has long been considered a sensitive window in which the environment may profoundly shape the brain and long-term health outcomes (Anderson & Thomason, 2013; Marzola et al., 2023). However, brain development does not unfold uniformly across gestation. Primary sensory and motor regions develop earlier in gestation than higher-order regions responsible for complex socioemotional and cognitive functioning. For instance, primary sensory and motor networks emerge by 20 weeks gestation (GA) and achieve an adult-like organization by birth, while higher-order networks appear later (e.g., a default-mode network component has been detected via fMRI at 35 weeks GA; Thomason et al., 2015) and remain relatively immature in utero (Jakab et al., 2014; Turk et al., 2015; Van den Heuvel et al., 2018). Both functional and structural connectivity also strengthen along medial-to-lateral and poster-to-anterior spatial axes (Chen et al., 2024; Van den Heuvel & Thomason, 2016). For example, myelination begins at 20 weeks GA in deep-brain subcortical structures and gradually extends from the posterior to frontal cortices over postnatal life (Lebel et al., 2007; Thomason, 2020; Yarnykh et al., 2018). Together, these spatiotemporal patterns of neuromaturation suggest that the entirety of gestation is not just one sensitive window for the entire brain. Instead, the timing of vulnerability may vary across regions and be determined by their unique developmental trajectories. Such region-specific windows have yet to be characterized in a large human cohort before birth. Defining prenatal sensitive windows would allow us to identify periods of both risk and opportunity. For instance, administering maternal prenatal interventions when the fetal brain is highly responsive to environmental input may maximize intergenerational benefit for pregnant parents and offspring alike. Knowledge of region-specific sensitive windows in gestation could also inform targeted and time-sensitive intervention after birth. For example, if we know that motor systems are most vulnerable to toxic exposures during later gestation, we could administer motor-focused interventions to infants who were exposed at that time, rather than waiting for observable symptoms to arise. This work may enhance both our etiological understanding of developmental psychopathology and the precision of early intervention. Given the difficulties of studying intrauterine development, much of our current understanding of prenatal sensitive windows stems from postnatal research. One body of work has examined how gestational timing moderates associations between exposure to known neurotoxicants, like prenatal stress, and infant neurodevelopment. Of note, natural experiments that rely on exposures to discrete disasters have found that prenatal exposure to maternal stress following a disaster predicts worse motor outcomes in infancy, but only when the disaster occurs later in pregnancy (Cao et al., 2014; Moss et al. 2017, Simcock et al., 2016). These timing effects may suggest that prenatal stress alters the development of motor brain regions, and that motor regions become increasingly vulnerable to stress exposure across gestation. In particular, later stress exposure may disproportionately impact the cerebellum, a highly-connected motor hub region that develops rapidly during the third trimester (Karolis et al., 2023; van den Heuvel et al., 2018). Given that hubs facilitate widespread network communication, environmental perturbation of hubs like the cerebellum can result in particularly strong functional deficits (van den Heuvel & Sporns, 2013). However, because these exposure studies did not directly measure brain development, the exact neural mechanisms and brain regions involved remain unclear. Furthermore, even disaster exposures may be confounded by prolonged stress and hazardous exposures (e.g., pollutants) that persist long after the disaster ends. These limitations highlight the challenges of defining sensitive windows in humans based on behavioral research, and the need to use fine-grained neuroimaging measures to directly index vulnerability within the fetal brain. Fetal neuroimaging has advanced significantly in recent years, with particular emphasis placed on understanding how functional connectivity and neural networks emerge in utero. Yet connectivity-based measures that depend on the existence of neural connections may fail to detect meaningful changes in local brain activity. This regional activation may be especially relevant to the fetal brain, which is dominated by weaker and shorter-range connections than those in infants and adults (Edde et al., 2019). In particular, prior work suggests that the magnitude of spontaneous brain activity may index regional maturation underlying sensitive window opening and closing (Herzberg et al., 2026; Molnár et al., 2021). Mapping local spontaneous activity across gestation may allow us to characterize sensitive windows for individual regions, even before neural connections are well-established. One proposed measure of spontaneous brain activity is the amplitude of low frequency fluctuations (ALFF) derived from the resting-state fMRI BOLD signal (Amador-Tejada et al., 2025). Unlike functional connectivity, which represents the “highways” that connect different brain regions, ALFF quantifies the magnitude of activity within each region (Lv et al., 2018). Studies in neonates and older children suggest that spontaneous activity measured via ALFF increases as sensitive windows open, and decreases as they close (Herzberg et al., 2026; Huang et al., 2023; Ji et al., 2023; Sydnor et al., 2023). Defining trajectories of ALFF before birth may similarly indicate the progression of sensitive windows across gestation. To our knowledge, only one study has examined age-related changes in ALFF before birth. In a sample of 82 fetal participants, (Liu et al., 2026) found that ALFF decreased in the thalamus and increased in the anterior cingulate cortex (ACC) between 26 and 39 gestational GA. Spontaneous thalamic activity is thought to pattern sensory networks before birth (Molnár et al., 2021). As a result, intrinsic thalamic activity may gradually decrease in favor of increasing thalamocortical connectivity and network synchronization across the third trimester of gestation (Taymourtash et al., 2023). The ACC contributes to both the salience and early infant default-mode networks (Gao et al., 2009), which are present in late gestation but develop much more gradually than sensory networks (Scheinost et al., 2024; Thomason et al., 2015). Increasing spontaneous activity within the ACC could indicate the maturational onset of these two neural networks. However, it remains unclear how ALFF changes before 26 weeks GA, whether age-related changes in fetal ALFF might also follow non-linear age trajectories, and whether observed increases and decreases in ALFF reflect a true sensitive window opening or closing. Spatiotemporal mapping of ALFF earlier in gestation and within a larger fetal cohort may shed further light on regional maturation—and vulnerability—during the prenatal period. The proposed project will leverage an existing fetal fMRI dataset to examine linear and non-linear age-related changes in ALFF throughout the fetal brain between 22 and 39 gestational weeks. Findings may provide preliminary insight into region-specific sensitive windows across the developing connectome and across gestation.
Institution: New York University
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7168035968