Maternal brain plasticity during pregnancy: an unresolved biological question

Pregnancy is associated with substantial neurobiological adaptations that prepare the maternal organism for the physiological and behavioural demands of motherhood. Human neuroimaging studies have demonstrated pronounced pregnancy-associated changes in maternal brain structure and function, particularly within regions governing social cognition, emotional processing, motivation, and maternal caregiving behaviour. A landmark prospective study comparing women before conception and after their first pregnancy reported reductions in grey matter volume across regions subserving social cognition; these reductions persisted for at least two years postpartum and accurately distinguished parous from nulliparous women1. Subsequent population-based neuroimaging work extended these observations by demonstrating that parity markers are associated with distinct structural brain signatures decades after childbirth, indicating that gestational neural adaptations leave enduring rather than merely transient traces2. More recently, dense-sampling precision neuroimaging tracking an individual from preconception through two years postpartum mapped the gestational trajectory of these changes at high temporal resolution, revealing progressive reductions in cortical grey matter volume and thickness alongside concomitant increases in white matter microstructural integrity3. The morphometric profile of these gestational adaptations closely parallels that of adolescence—a second developmental window of intensive steroid-driven cortical reorganisation6—and extends to subcortical reward circuitry, where volumetric reductions in the ventral striatum predict postpartum maternal neural responsiveness to infant cues7. Longitudinal postpartum investigations further indicate that several of these structural alterations evolve non-linearly, with gestational grey matter reductions in certain regions followed by partial volumetric recovery or expansion across cortical and subcortical networks during the early postpartum weeks8,9, though the extent to which specific adaptations persist or reverse over subsequent years remains an active question10. Collectively, these findings establish that pregnancy entails dynamic, functionally relevant neural reorganisation linked to the transition to caregiving rather than passive exposure to endocrine fluctuations11,12.

Current models of maternal brain plasticity have centred largely on maternal ovarian hormones and classical neuroendocrine pathways; while these models account for major features of maternal neuroadaptation, they remain incomplete. Gestational surges in oestradiol, progesterone, prolactin, oxytocin, and glucocorticoids modulate neural circuits governing stress reactivity, reward processing, and maternal responsiveness, providing an essential foundation for understanding pregnancy-associated neuroadaptation13,14 and shaping female brain health across the lifespan15. However, gestation also involves sustained biochemical communication from the maternal-foetal interface. The placenta functions as a high-output transient endocrine organ that releases steroid and peptide hormones, growth factors, immune mediators, and extracellular vesicles into the maternal circulation. These placenta-derived signals orchestrate maternal metabolic, cardiovascular, immunological, and endocrine adaptations throughout gestation4,5, while parallel cross-talk across the same interface shapes foetal neurodevelopment and metabolic programming16. Despite its established role in governing maternal systemic physiology, the potential contribution of placenta-derived signalling to maternal brain plasticity remains insufficiently integrated into contemporary models of pregnancy neurobiology—an omission recently highlighted as a priority for longitudinal, multi-system investigation17.

This article provides a critical narrative review of the evidence addressing a central biological question: whether placenta-derived signals contribute to the physiological environment supporting maternal brain adaptation during pregnancy. It is structured as a critical review rather than a purely conceptual perspective because direct in vivo evidence in human pregnancy remains too limited to sustain an uncritical hypothesis-driven model. Four candidate pathways are examined sequentially: placental endocrine signalling, extracellular vesicle-mediated communication, immune modulation, and metabolic-vascular regulation. Each pathway is formulated as a mechanistic hypothesis and appraised across a standardised evidentiary sequence: biological rationale, confirmation of placental origin, evidence of maternal systemic action, direct evidence within the maternal brain, indirect or extrapolated evidence from foetal, animal, or pathological contexts, and remaining methodological uncertainties. Table 1 summarises this evidence appraisal across the four candidate pathways, and Figure 2 illustrates the anatomical routes by which placenta-derived signals could access the maternal central nervous system (CNS). The integrative placenta–maternal brain framework presented in Figure 1 and discussed in the closing sections thus emerges as a testable synthesis that complements, rather than displaces, established maternal neuroendocrine mechanisms.

Review methodology and literature selection

Literature for this review was identified through structured searches of PubMed/MEDLINE, Scopus, and Web of Science, supplemented by Google Scholar and backward and forward citation tracking of seminal articles, covering the period from January 1990 to June 2026 without language restrictions at initial screening. Search queries combined three concept blocks using Boolean operators. The first block targeted the maternal brain using the terms "maternal brain", "matrescence", and "pregnancy" paired with "neuroplasticity", "grey matter", "gray matter", "neuroimaging", or "maternal behaviour". The second block targeted the placenta and its secretory products using the terms "placenta", "syncytiotrophoblast", "placental hormone", "placental lactogen", "chorionic gonadotropin", "placental corticotropin-releasing hormone", "allopregnanolone", "extracellular vesicle", "exosome", and "C19MC". The third block targeted candidate CNS access and communication pathways using the terms "blood-brain barrier", "circumventricular organ", "transcytosis", "neuroimmune", "cytokine", "nutrient sensing", and "cerebral haemodynamics". Reference lists of retrieved review articles were manually screened to capture additional primary studies not indexed by the database queries.

Articles were selected based on their evidentiary contribution to one or more of the four mechanistic hypotheses rather than for exhaustive coverage of each subdiscipline, with priority given to primary human studies, mechanistic animal investigations where human data were unavailable, and authoritative reviews synthesising mature physiological literature. Records were retained if they established a placental origin for a signalling molecule, demonstrated a maternal systemic effect of that molecule, or characterised an action of that molecule (or its molecular class) on neural tissue; purely clinical obstetric outcome studies lacking a signalling or neurobiological component were excluded. Consistent with a critical narrative review design, no formal risk-of-bias scoring instrument or quantitative meta-analysis was applied; literature screening and study selection were conducted by the first author, independently verified by the co-authors, and reconciled through consensus discussion. To maintain strict evidentiary transparency, every mechanistic claim in the subsequent sections is explicitly graded according to three predefined levels of evidence:

• Direct evidence: Empirical observation within maternal neural tissue or in the maternal brain in vivo.

• Indirect evidence: Demonstration of a maternal systemic physiological effect whose downstream neural consequence was not directly measured.

• Extrapolated evidence: Inference derived from the foetal compartment, non-human experimental models with divergent placentation, or pathological pregnancy states (such as pre-eclampsia).

These same three evidentiary tiers structure the comparative appraisal presented in Table 1.

Beyond maternal hormones: expanding the endocrine framework of maternal brain adaptation

Maternal brain plasticity during pregnancy has traditionally been interpreted through the actions of hormones synthesised by maternal endocrine glands—a framework that remains well supported but physiologically incomplete. Circulating concentrations of oestradiol, progesterone, prolactin, oxytocin, and glucocorticoids undergo marked gestational fluctuations and modulate neural circuits governing stress responsiveness, social cognition, reward motivation, and maternal caregiving13,14. Extensive preclinical and clinical evidence confirms that these endocrine adaptations are central to gestational neurobiology, with circulating hormones binding to cognate receptors across hypothalamic, limbic, and cortical networks to remodel neuronal excitability, synaptic architecture, and behavioural output18,19.

Nevertheless, the endocrine milieu of pregnancy extends beyond maternal glandular secretion to encompass a transient, anatomically distinct feto-placental source. Throughout gestation, the placental syncytiotrophoblast secretes substantial quantities of steroid hormones, peptide hormones, growth factors, and bioactive signalling molecules directly into the maternal circulation, coordinating the metabolic, cardiovascular, immunological, and reproductive adaptations required to sustain pregnancy4,5,20. Crucially, the placenta functions not merely as a constitutive secretory conduit, but as an active metabolic sensor that dynamically adjusts its growth, nutrient transport capacity, and endocrine output in response to maternal nutrient and oxygen availability21,22. Because placental endocrine secretion varies continuously with maternal metabolic and nutritional status, any downstream placental influence on maternal neuroplasticity would likewise be expected to operate as a context-dependent, dynamic signal rather than a fixed gestational programme.

The most compelling endocrine candidate linking placental function to maternal neural adaptation is placental steroidogenesis. During the second and third trimesters of human pregnancy, the placenta becomes the dominant source of circulating progesterone and oestrogens, converting maternal and foetal steroid precursors in large quantities. Progesterone is metabolised both peripherally and within the CNS into the neuroactive steroid allopregnanolone, a potent positive allosteric modulator of gamma-aminobutyric acid type A (GABAA) receptors23. Preclinical rodent studies provide direct neural evidence that the gestational rise and peripartum withdrawal of brain allopregnanolone drive structural and functional reorganisation of GABAA receptor subunit composition in the maternal brain24. In human pregnancy, lower mid-gestational circulating allopregnanolone concentrations have been prospectively associated with postpartum depressive symptoms, and the abrupt loss of placental progesterone synthesis at parturition provides a coherent mechanistic explanation for the acute postpartum drop in central neurosteroid tone25. Because the human placenta supplies the overwhelming majority of the progesterone substrate from which allopregnanolone is synthesised during the second half of gestation, placental steroidogenic output currently represents the strongest candidate for a placenta-dependent modulator of maternal neural function.

Two additional placenta-derived peptide hormones warrant classification as primary candidates for central maternal signalling because both possess verified receptor targets within the mammalian brain: placental corticotropin-releasing hormone (pCRH) and human chorionic gonadotropin (hCG). Anthropoid primates, including humans, are unique among eutherian mammals in that the syncytiotrophoblast secretes pCRH into the maternal circulation at exponentially rising concentrations across gestation, progressively recalibrating the maternal hypothalamic-pituitary-adrenal (HPA) axis and stress responsiveness26; moreover, accelerated mid-gestational pCRH trajectories have been linked prospectively to postpartum depressive symptomatology27. Similarly, hCG binds to functional luteinizing hormone/chorionic gonadotropin (LH/hCG) receptors expressed in multiple neural regions, including the hippocampus and hypothalamus28. In cycling female rats, peripheral and intracerebroventricular administration of hCG crosses the blood-brain barrier in small but intact fractions and alters hippocampus-dependent exploratory and neophobic behaviours29, while circulating maternal hCG concentrations in human gestation vary systematically with foetal sex and maternal physiological characteristics30. Both pCRH and hCG therefore satisfy the criteria of confirmed placental origin, established maternal systemic action, and plausible central receptor targets, setting them apart from placental factors whose neurobiological relevance relies solely on systemic inference.

Several additional placenta-derived hormones exhibit biological properties pertinent to maternal neurobiology, although the supporting evidence remains more indirect. Human placental lactogen (hPL) and placental growth hormone (hPGH) regulate maternal insulin sensitivity, lipolysis, and nutrient partitioning—processes increasingly recognised as modulators of central energy homeostasis and behaviour5,31. In rodents, lactogenic signalling exerts direct central actions: activation of central prolactin receptors stimulates forebrain neurogenesis during gestation32, and circulating lactogens enter the rodent brain via a saturable transport mechanism that operates independently of the prolactin receptor itself33. However, whether placenta-derived hPL, as distinct from maternal pituitary prolactin, crosses the human blood-brain barrier to engage maternal neural circuits remains unverified in vivo, rendering this extrapolated evidence. Similarly, placental modulation of maternal immune and vascular pathways may shape neural adaptation indirectly by altering the systemic physiological environment in which gestational neuroplasticity unfolds34,35. On the foetal side of the interface, conditional genetic ablation of placental allopregnanolone synthesis in mice disrupts cerebellar white matter myelination and induces autism-like social deficits in male offspring36, providing definitive proof-of-principle that placental endocrine output can sculpt mammalian brain development while leaving open whether the adult maternal brain exhibits comparable sensitivity. Furthermore, because placental hormone secretion varies across gestational age and differs according to foetal sex20, the maternal systemic milieu supporting neuroadaptation is unlikely to be uniform across pregnancies.

Before appraising the four individual pathways in detail, it is essential to distinguish between two conceptual accounts of how placental activity could relate to maternal brain adaptation. These accounts are not mutually exclusive, but they entail markedly different evidentiary thresholds:

1. Active instructive signalling: In this model, placenta-specific molecules engage defined receptors or cellular targets within maternal neural circuits to direct neurochemical, structural, or behavioural adaptations (e.g., priming caregiving motivation or affective regulation). Placental neurosteroid precursors (via allopregnanolone), pCRH, and hCG represent the primary candidates for instructive signalling because each possesses identified molecular targets in neural tissue.

2. Allostatic or homeostatic remodelling: In this model, the maternal brain undergoes structural and metabolic reorganisation as a secondary adaptation to the substantial haemodynamic, immunological, and metabolic workload imposed by the feto-placental unit. Here, neural plasticity reflects central homeostatic accommodation to systemic physiological demand rather than purposive molecular instruction from the placenta.

Much of the empirical literature appraised below is fully compatible with allostatic remodelling. Maintaining a rigorous distinction between instructive signalling and allostatic adaptation prevents the misinterpretation of secondary systemic adjustments as directed feto-placental programming of the maternal brain. Establishing active instructive signalling requires demonstrating that a placenta-specific ligand accesses a defined maternal neural target and that selective perturbation of that ligand alters maternal neural or behavioural outcomes—an evidentiary standard not yet met by any placental signal in human pregnancy.

An expanded neuroendocrine framework of pregnancy should therefore integrate both maternal glandular hormones and placenta-derived signals without attributing greater explanatory weight to placental communication than current evidence warrants. Reviews of female brain ageing have similarly proposed that the profound immunoendocrine transitions of pregnancy, alongside menopause, shape long-term trajectories of neurological health and neurodegenerative risk37. Defining how placenta-derived signals interact with established maternal hormonal pathways is the critical step required to transform this conceptual framework into a falsifiable neurobiological model.

Candidate mechanistic pathways and the levels of evidence supporting them

Determining whether placenta-derived signals contribute to maternal brain adaptation requires specifying the biological pathways linking the maternal-foetal interface to maternal physiological systems and grading the empirical evidence supporting each step. Throughout gestation, the syncytiotrophoblast releases steroid and peptide hormones, extracellular vesicles, immunomodulatory cytokines, and metabolic-vascular regulators into the maternal circulation4,5. Below, four candidate pathways are formulated as mechanistic hypotheses and evaluated across a uniform evidentiary sequence: biological rationale, placental origin, maternal systemic effect, direct evidence in the maternal brain, indirect or extrapolated evidence, and remaining uncertainties. Table 1 synthesises this pathway-by-pathway appraisal.

Distinguishing placental signals from maternal background: methodological requirements

The principal technical obstacle to testing any of the four hypotheses in human pregnancy is biochemical overlap: because placenta-derived and maternal endogenous molecules circulate together in maternal blood, observing a statistical correlation between a circulating hormone or cytokine and a neuroimaging metric does not establish the tissue of origin. Three molecular profiling strategies now offer the analytical specificity required to overcome this limitation in human cohort studies:

1. Placenta-specific microRNA signatures: Targeting nucleic acids expressed almost exclusively by trophoblast cells—specifically the primate-specific C19MC microRNA cluster exported within exosomes (and the orthologous chromosome 14 microRNA cluster [C14MC] in rodents for comparative preclinical studies)—provides an unambiguous molecular tag of placental origin in maternal plasma42.

2. Immunoaffinity isolation of syncytiotrophoblast EVs: Capturing PLAP-positive extracellular vesicles from maternal peripheral blood via immunoaffinity chromatography or fluorescence nanoparticle tracking physically separates syncytiotrophoblast-derived vesicles from the much larger pool of maternal platelet-, endothelial-, and leukocyte-derived EVs prior to proteomic or transcriptomic cargo analysis41.

3. Maternal plasma cell-free RNA (cfRNA) profiling: Quantifying trophoblast-specific gene transcripts within maternal circulating cfRNA already enables accurate non-invasive prediction of gestational age, preterm birth, and pre-eclampsia58,59. Pairing longitudinal cfRNA quantification of placental transcripts with serial structural and functional neuroimaging in the same pregnant participants would allow investigators to test whether fluctuations in placenta-specific transcriptional output covary with maternal cortical and subcortical remodelling while statistically controlling for maternal systemic background.

Complementing human longitudinal studies, experimental animal and ex vivo models remain indispensable for establishing causal mechanisms that cannot be ethically manipulated in pregnant women. However, pronounced interspecies divergence in placental anatomy and endocrinology dictates that experimental model selection determines translational validity. Laboratory rodents possess haemochorial placentation but express expanded families of prolactin-like proteins lacking human orthologues, rely on the maternal corpus luteum rather than the placenta for progesterone synthesis throughout most of gestation, and do not express placental CRH60,61,62. Furthermore, rodent maternal caregiving behaviour is heavily dependent on olfactory processing and the medial preoptic area, whereas human maternal adaptation engages distributed higher-order cortical networks subserving theory of mind and social cognition12,19. Consequently, experimental designs should prioritise model systems that preserve the relevant human biology:

• Non-human primates: Anthropoid primates share both haemochorial villous placentation and exponentially rising syncytiotrophoblast pCRH secretion across gestation26,60.

• Trophoblast-specific genetic manipulation in mice: Conditional gene targeting using trophoblast-restricted Cre recombinase drivers (such as Cyp19a1-Cre)63 or blastocyst trophectoderm-specific lentiviral gene transfer prior to implantation64 enables selective deletion or overexpression of a placental gene while leaving the maternal genome—and maternal endogenous endocrine/immune tissues—completely intact.

• Ex vivo dual perfusion of the human placental cotyledon: The dually perfused human term placental lobule allows the directional release of hormones, cytokines, and PLAP-positive EVs into the maternal-side perfusate to be quantified in human tissue in the complete absence of maternal systemic confounding65.

The power of trophoblast-restricted genetic manipulation is exemplified by the conditional deletion of the allopregnanolone-synthesising enzyme Akr1c14 specifically within mouse trophoblast (Cyp19a1-Cre), which definitively demonstrated that placental neurosteroid output causally regulates offspring cerebellar myelination and social behaviour36. Executing the analogous placenta-specific knockout experiment while evaluating maternal brain structure, microglial morphology, GABAA receptor plasticity, and postpartum caregiving behaviour in the genetically wild-type dam represents the single most informative preclinical experiment needed to test the placenta–maternal brain hypothesis.

Defining the placenta-maternal brain hypothesis: priorities for future research

Determining whether placenta-derived signals actively shape maternal neuroplasticity requires research designs that bridge reproductive biology and maternal systems neuroscience. As our evidentiary appraisal demonstrates, biological plausibility is strongly supported by the placenta's established governance of maternal systemic physiology, whereas direct empirical proof linking placenta-derived molecules to maternal neural circuits remains largely confined to neuroactive steroids and preclinical rodent models. Future investigations should therefore move beyond isolated single-hormone correlations and examine gestation as an integrated multi-organ communication network encompassing the placenta, maternal endocrine axes, peripheral immunity, cerebrovascular autoregulation, and the CNS17.

Longitudinal human cohort studies pairing high-frequency placental molecular profiling with precision maternal neuroimaging represent the highest-priority translational step. Recent dense-sampling neuroimaging work has demonstrated that maternal cortical volume, thickness, and white matter quantitative anisotropy can be tracked week-by-week from preconception through the postpartum period3. Integrating this precision imaging paradigm with serial maternal blood sampling for PLAP-positive EV cargo, C19MC microRNAs, trophoblast-specific cfRNA transcripts, pCRH, and neurosteroid metabolomics would test whether individual trajectories of placental signalling predict the magnitude or regional specificity of maternal brain remodelling. Crucially, incorporating the placenta-specific molecular assays detailed in Section 5 is what will allow investigators to attribute observed brain–biomarker associations to placental output rather than maternal glandular activity.

Human cohort designs should also measure peripheral immune, metabolic, cardiovascular, and psychological variables concurrently with placental biomarkers, as the biologically relevant question is how placental signals interact with maternal endogenous pathways rather than whether they act in isolation. Stratification by foetal sex is particularly essential in these cohorts: first-trimester and term human placental transcriptomes exhibit pronounced sexual dimorphism66, maternal peripheral blood mononuclear cell cytokine responses across gestation vary systematically with foetal sex67, and maternal circulating hCG trajectories differ between male- and female-bearing pregnancies30. Testing whether foetal sex modifies gestational maternal brain plasticity—or alters maternal vulnerability to peripartum mood disorders—will require large, demographically diverse longitudinal cohorts12.

Finally, investigations of extracellular vesicle and neuroimmune pathways must rigorously separate the physiological signalling of uncomplicated gestation from the pathophysiological cascades of obstetric syndromes such as pre-eclampsia. In healthy pregnancy, circulating PLAP-positive EVs rise gradually and support a controlled, adaptive priming of maternal innate immunity and vascular compliance43,44. In pre-eclampsia, by contrast, hypoxic and oxidatively stressed syncytiotrophoblast tissue sheds elevated quantities of EVs enriched in anti-angiogenic factors, pro-inflammatory cytokines, and oxidized lipids that provoke systemic endothelial dysfunction, hypertension, and blood-brain barrier breakdown [45,46,48]. Because much of the experimental literature on placental EVs and cerebrovascular barrier function has been generated in the context of pre-eclampsia, care must be taken not to conflate pathological barrier injury with physiological feto-maternal neurocommunication. Studies aiming to define normative maternal brain adaptation must establish uncomplicated gestation as their primary mechanistic reference.

An integrative, placenta-inclusive framework does not replace classical neuroendocrine models of the maternal brain; rather, it embeds them within the complete reproductive physiology of pregnancy, treating the maternal-foetal interface as a candidate co-regulator of the internal milieu supporting maternal neuroplasticity while maintaining a strict distinction between instructive placental signalling and allostatic maternal remodelling.

Conclusion: towards an integrated model of maternal brain adaptation

Maternal brain plasticity during pregnancy is a robust, reproducible neurobiological phenomenon, yet the upstream physiological signals that initiate and coordinate this remodelling remain incompletely defined. While classical models have rightly emphasised maternal ovarian steroids and pituitary neuroendocrine pathways, gestation also involves continuous, high-output biochemical signalling from the placenta—a transient endocrine organ that reshapes maternal physiology through steroidogenesis, peptide hormone secretion, nutrient sensing, immunomodulation, and extracellular vesicle release.

Our critical appraisal of the literature indicates that current evidence does not yet prove that placenta-derived signals directly instruct maternal neural circuits in human pregnancy, and shows that the four candidate pathways differ substantially in their evidentiary proximity to that standard. Placental steroidogenic output holds the strongest mechanistic claim via progesterone-derived allopregnanolone, supported by direct neural remodelling evidence in rodents and prospective clinical associations in women, followed by pCRH and hCG, which engage verified central receptors but lack direct neuroimaging correlates in human pregnancy. Extracellular vesicle, neuroimmune, and metabolic-vascular pathways currently rest on indirect systemic and extrapolated pathological or foetal evidence. Integrating placental biology into models of maternal neuroplasticity provides a rigorous, biologically complete framework that complements precision human neuroimaging3,17 and generates falsifiable predictions for multidisciplinary research across reproductive biology, endocrinology, and neuroscience.

Declarations