Maternal Anxiety During Pregnancy and Offspring Information Processing Across Development: A Theory-Guided Predictive-Processing Reappraisal of Two Longitudinal Cohorts
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Genomics Radio. Generated commentary on the latest computational biology and genomics papers.
Ines: We're looking at the paper: "Maternal Anxiety During Pregnancy and Offspring Information Processing Across Development".
Marcus: Here's the gist of the paper. Maternal anxiety during pregnancy is associated with selective differences in sensory, affective, attentional, cognitive, neural, and interoceptive-affective processing across development <ref:2608.24983#pg2> The findings are compatible with a parsimonious PP-informed account in which maternal anxiety during pregnancy is associated with differences in how relevance and processing resources are allocated to predictable, threat-related, affectively ambiguous, contextual, feedback-related, and bodily information <ref:2608.24983#pg8> This empirical reappraisal shows that findings from the two longitudinal cohorts form a selective, multimodally convergent pattern that PP can organize more precisely than generalized accounts of impairment or heightened reactivity <ref:2608.24983#pg8> Pruning redundant reasoning tokens The study re-examines findings across eleven interrelated publications from two non-clinical cohorts to assess whether they form a coherent PP-compatible developmental pattern <ref:2608.24983#pg2> The research examines whether differences originally described across sensory, affective, cognitive, neural, and interoceptive domains can be organized in terms of the allocation of relevance and processing resources and the development of internally organized regulation <ref:2608.24983#pg2> This cross-publication synthesis could not be provided by any individual report and offers a longitudinal complement to newer studies that test particular PP mechanisms more directly but over narrower developmental periods <ref:2608.24983#pg2> The research focuses on the Leuven and Tilburg cohorts because, to our knowledge, few if any prospective cohorts combine comparably dense longitudinal phenotyping with a comparable developmental span <ref:2608.24983#pg2> How the training signal is built The paper proposes that perception, cognition, and action arise from interactions between incoming information and internally generated predictions partly governed by precision <ref:2608.24983#pg2> Precision refers to this context-dependent weighting and regulates how strongly sensory signals or prediction errors influence updating and action <ref:2608.24983#pg2> The authors examine a parsimonious developmental hypothesis that variation in maternal anxiety during pregnancy is associated with differences in how the developing brain allocates relevance and processing resources to sensory, affective, contextual, feedback-related, and bodily information <ref:2608.24983#pg2> Within PP accounts, their context-dependent influence may be mediated by precision weighting <ref:2608.24983#pg2> Where the gains come from The findings from the Leuven cohort indicate that prenatal exposure to higher maternal anxiety was associated with later self-regulatory difficulties and symptoms relevant to developmental psychopathology <ref:2608.24983#pg8> Differences already appeared before birth, with maternal anxiety during weeks 12-22 being associated with altered fetal behavioural-state organization <ref:2608.24983#pg8> During infancy, exposed offspring showed more regulatory difficulties <ref:2608.24983#pg8> In adolescence, prenatal maternal anxiety was additionally associated with HPA-axis dysregulation in male and female offspring <ref:2608.24983#pg8> Developmental Integration Across Cohorts and Possible Pathways to Later Vulnerability The Leuven cohort identified selective differences in internally organized cognitive control and its neural correlates from adolescence into adulthood <ref:2608.24983#pg2> The Tilburg cohort examined earlier sensory, affective, and attentional processing during infancy and early childhood <ref:2608.24983#pg2> One possible developmental interpretation is that early differences in sensory and affective weighting contribute to the calibration of broader regulatory systems <ref:2608.24983#pg8> Persistent processing of repetitive input, increased weighting of threat-related vocal information, or enhanced attention to affectively uncertain neutral stimuli could increase the amount of information treated as potentially relevant <ref:2608.24983#pg8> Interpretational Constraints and Falsifiability Considerations A first constraint concerns evidential independence, which limits the strength of any cross-publication inference <ref:2608.24983#pg8> A second constraint concerns the specificity of PP constructs, as none of the original publications formally estimated computational precision, learning rates, uncertainty, volatility, or directional prediction-error signalling <ref:2608.24983#pg8> The present findings do not confirm or falsify PP as a general framework but limit the interpretations that can plausibly be derived from it <ref:2608.24983#pg8> Future research should test these interpretations more directly by manipulating predictability, uncertainty, volatility, and sensory reliability <ref:2608.24983#pg8> Taken together, the findings are compatible with a parsimonious PP-informed account in which maternal anxiety during pregnancy is associated with differences in how relevance and processing resources are allocated to predictable, threat-related, affectively ambiguous, contextual, feedback-related, and bodily information and in how regulation becomes internally organized <ref:2608.24983#pg8> The strongest cross-cohort conclusion is not that prenatal maternal anxiety produces a single fixed PP phenotype but rather both cohorts reveal selective developmental differences in the regulation of information relevance <ref:2608.24983#pg8> The value of these findings, when interpreted from a PP perspective, lies in distinguishing among competing PP-compatible explanations rather than in demonstrating the framework itself <ref:2608.24983#pg8> Declarations Data availability.
Ines: This article reinterprets previously published data from the Tilburg Prenatal Early Life Stress cohort and the Leuven Prenatal Project; no new data were generated <ref:2608.24983#pg8> Funding.
Marcus: The PELS study was supported by the national funding agencies participating in the European Science Foundation EUROCORES Programme EuroSTRESS (EuroSTRESS-PELS, grant 99930AB6-0CAC-423B-9527-7487B33085F3), including the Brain and Cognition Programme of the Netherlands Organisation for Scientific Research (NWO) <ref:2608.24983#pg8> Both the PELS and Leuven cohorts were partly supported by the European Commission Seventh Framework Programme (FP7-HEALTH-2011.2.2.2-2, BRAINAGE, grant agreement 279281) <ref:2608.24983#pg8> The Leuven cohort was additionally supported by the Research Foundation - Flanders (FWO-Vlaanderen; projects 1197285N and 1197287N) <ref:2608.24983#pg8> BRHVdB was further supported by FWO-Vlaanderen SBO project S003524N and by COST Action CA22114, funded by the European Cooperation in Science and Technology <ref:2608.24983#pg8> The authors used ChatGPT and Claude to assist with literature organization, drafting, and language editing during the preparation of this work <ref:2608.
Title and authors: Ines: Let's talk about the title again. Maternal Anxiety During Pregnancy and Offspring Information Processing Across Development: A Theory-Guided Predictive-Processing Reappraisal of Two Longitudinal Cohorts. It really tells you that this isn't just a study on anxiety; it’s a study on how that anxiety shapes the very machinery of how the brain processes information over time.
Marcus: I agree with Ines there. And when you see "theory-guided reappraisal," it signals that they aren't just describing observations; they are trying to fit those observations into a specific computational framework, which is predictive processing.
Yuki: From my side, the focus on two longitudinal cohorts—Leuven and Tilburg—is what makes this interesting for the wider field. It’s about seeing if these selective differences hold up across different developmental windows and populations.
Ines: That's a good point, Yuki. The paper is trying to show that when you look at how they organized the findings across eleven related publications, there's this coherent pattern that predictive processing can explain better than just saying people are generally impaired or more reactive.
Marcus: So, the authors are essentially using this framework to organize what was previously seen as disparate findings from different studies into a single developmental story about information allocation.
The paper's summary: Ines: The summary of the paper explains that prenatal distress has been linked to changes in brain and behavior, but the literature hasn't really framed it computationally before. So, this work reappraises those findings across eleven related studies to see if they fit a consistent predictive processing model.
Marcus: And what they found is that when you look at task goals or stimulus relevance needing to be generated internally—not just responding to external cues—that’s where the differences in offspring emerge.
Yuki: It sounds like the key finding is that prenatal maternal distress might influence how the developing brain sets up its internal models for what matters and what doesn't matter as a child.
Ines: That's right. And they propose that this allocation of resources is split between different types of information, specifically predictable things, threat-related things, ambiguous things, contextual things, feedback-related things, and bodily information.
Marcus: They are suggesting that the prenatal anxiety affects how those different categories get their processing power allocated across development.
The paper's improvements: Ines: Now for the part where they suggest improvements or a new way to look at things. They propose that perception, cognition, and action come from interactions between incoming information and predictions that are partly governed by precision weighting.
Marcus: Precision weighting is the mechanism here, which they say regulates how strongly sensory signals or prediction errors influence updating and action. It’s this context-dependent weighting that they think mediates the prenatal anxiety effects within a predictive processing account.
Yuki: From a population perspective, if we can map where those differences in relevance allocation start—like with sensory or affective weighting during infancy, as seen in the Tilburg data—it gives us a pathway to predict later socio-emotional difficulties.
Ines: That’s the developmental trajectory prediction aspect. The paper suggests that persistent processing of repetitive input, increased weighting of threat related vocal information, or enhanced attention to affectively uncertain neutral stimuli could increase the amount of information treated as potentially relevant.
Conclusion: Marcus: So to wrap up on this Maternal Anxiety During Pregnancy and Offspring Information Processing Across Development paper, it seems the strongest conclusion isn't that prenatal anxiety creates one single fixed pattern in the offspring, but rather that both cohorts show selective developmental differences in how information relevance gets regulated.
Ines: That’s a crucial distinction. It’s not about a universal effect across all development; it’s about specific windows where the regulation of information relevance is altered.
Yuki: And for the wider picture, this work helps us distinguish between different predictive processing explanations because it shows how these developmental differences manifest across the spectrum of information types.
Marcus: So, we've seen how they suggest that prenatal maternal anxiety influences how relevance and processing resources are allocated to predictable, threat related, ambiguous, contextual, feedback related, and bodily information.
Ines: That’s the core idea: it's about how those different kinds of input get weighted differently as the brain matures.
Yuki: It opens up a way for future work to test these specific mechanisms more directly by manipulating predictability and uncertainty in ways that correspond to what they found here.
Marcus: Yeah, so we've heard about the predictive processing framework and how it organizes the prenatal anxiety findings from this paper.
Research Group Health Psychology and Leuven Brain Institute, KU Leuven (University of Leuven) · Institute on Human Development and Disability, School of Medicine, University of Washington
q-bio.QM, q-bio.NC
Submitted: 2026-08-25
Updated: 2026-11-03
Comments: 50 pages, 5 tables, 1 figure; supplementary material (Figure S1, Tables S1-S4) included. v3: STAI-State cutoff definition clarified; cross-cohort comparison of anxiety measures revised; Tables S2-S4 moved to the supplementary material; highlights revised. Companion theory paper https://doi.org/10.31234/osf.io/zj4bf_v1. Final author-approved version: polishing touches
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
Importance score: 90/100
The gist: Maternal anxiety during pregnancy is associated with selective differences in sensory, affective, attentional, cognitive, neural…
Key concepts
- Precision Weighting
- This refers to a context-dependent way the brain weighs incoming sensory signals or prediction errors. It determines how strongly these signals influence updating and subsequent actions. High precision weighting means certain inputs are weighted more heavily in specific situations.
- Information Allocation
- This concept describes how the developing brain chooses which information to focus on and process based on its current needs. Maternal anxiety is hypothesized to selectively allocate resources toward predictable, threat-related, or bodily information over other types of input.
- Internal Regulation
- This refers to how an individual develops their own systems for self-control and managing their responses. The study suggests that early differences in sensory and affective weighting calibrate these broader regulatory systems as individuals develop over time.
Terminology
Summary
Maternal anxiety during pregnancy is associated with selective differences in sensory, affective, attentional, cognitive, neural, and interoceptive-affective processing across development <ref:2608. p. 8 Aim and Research Question
We ask whether findings across eleven interrelated publications from the Leuven and Tilburg
cohorts form a selective developmental pattern associated with maternal anxiety during
pregnancy. Specifically, we examine whether differences originally described across sensory,
affective, cognitive, neural, and interoceptive domains can be organized in terms of the
allocation of relevance and processing resources and the development of internally organized
regulation. 3
allocated to predictable, threat-related, ambiguous, contextual, feedback-related, and bodily
information. Because the publications did not formally estimate precision, prediction errors, or
priors, we interpret them as hypothesis-supporting evidence for a PP-informed developmental
account rather than a formal test. Predictive-processing (PP)
accounts propose that perception, cognition, and action arise from interactions between incoming
information and internally generated predictions, partly governed by precision. We reappraised
findings across eleven interrelated publications from two non-clinical cohorts to assess whether
they form a coherent PP-compatible developmental pattern.Where the paper says it
Where the paper says it
Where the paper says it
Pruning redundant reasoning tokens
The study re-examines findings across eleven interrelated publications from two non-clinical cohorts to assess whether they form a coherent PP-compatible developmental pattern <ref:2608.24983#pg2> The research examines whether differences originally described across sensory, affective, cognitive, neural, and interoceptive domains can be organized in terms of the allocation of relevance and processing resources and the development of internally organized regulation <ref:2608. p. 8 Aim and Research Question
We ask whether findings across eleven interrelated publications from the Leuven and Tilburg
cohorts form a selective developmental pattern associated with maternal anxiety during
pregnancy. Specifically, we examine whether differences originally described across sensory,
affective, cognitive, neural, and interoceptive domains can be organized in terms of the
allocation of relevance and processing resources and the development of internally organized
regulation. 7
use their developmental and multimodal complementarity to ask whether selective findings from
infancy to adulthood can be organized around common questions: which information is treated
as relevant, how task goals and expected outcomes are maintained, how feedback and context
influence action, and how bodily signals acquire affective meaning. This cross-publication
synthesis could not be provided by any individual report and offers a longitudinal complement to
newer studies that test particular PP mechanisms more directly but over narrower developmental
periods. This cross-publication
synthesis could not be provided by any individual report and offers a longitudinal complement to
newer studies that test particular PP mechanisms more directly but over narrower developmental
periods. This reappraisal does not attempt exhaustive coverage of the broader prenatal-distress
neuroimaging literature cited above; we focus on the Leuven and Tilburg cohorts because, to our
knowledge, few if any prospective cohorts combine comparably dense longitudinal phenotyping
with a comparable developmental span, the Leuven cohort was, internationally, among the first
to include prenatal assessment and to follow the same cohort, modest in size, into adulthood.Where the paper says it
Where the paper says it
Where the paper says it
How the training signal is built
The paper proposes that perception, cognition, and action arise from interactions between incoming information and internally generated predictions partly governed by precision <ref:2608.24983#pg2> Precision refers to this context-dependent weighting and regulates how strongly sensory signals or prediction errors influence updating and action <ref:2608. p. 4 5
by prior beliefs, or priors, while their influence relative to incoming information depends on
estimated reliability. Precision refers to this context-dependent weighting and regulates how
strongly sensory signals or prediction errors influence updating and action. , 2014, 2017), while computational work suggests that such differences may be task- and
context-dependent rather than reflecting a generic prediction deficit (Arthur et al. We examine a parsimonious developmental hypothesis: that variation in maternal anxiety during
pregnancy is associated with differences in how the developing brain allocates relevance and
processing resources to sensory, affective, contextual, feedback-related, and bodily information,
and in how cognitive and behavioural regulation becomes internally organized. We use
“relevance” and “processing-resource allocation” as functional descriptions of the empirical
pattern. Within PP accounts, their context-dependent influence may be mediated by precision
weighting, but the original publications did not estimate this mechanism directly.Where the paper says it
Where the paper says it
Where the paper says it
Where the gains come from
The findings from the Leuven cohort indicate that prenatal exposure to higher maternal anxiety was associated with later self-regulatory difficulties and symptoms relevant to developmental psychopathology <ref:2608.24983#pg8> Differences already appeared before birth, with maternal anxiety during weeks 12-22 being associated with altered fetal behavioural-state organization <ref:2608. p. 1 2
Keywords
Maternal anxiety; Predictive processing; Precision weighting; Cognitive control; Event-related
potentials; Interoception; Neurodevelopment
Abstract (word count: n= 250)
Prenatal maternal distress has been associated with differences in offspring brain and behavioural
development, yet this literature is rarely framed computationally. Predictive-processing (PP)
accounts propose that perception, cognition, and action arise from interactions between incoming
information and internally generated predictions, partly governed by precision. 3 in the high-anxiety group. Depending on the analytic follow-up
sample, approximately 25-30% of offspring were classified as prenatally exposed to higher
maternal anxiety. Prenatal maternal psychological distress has
repeatedly been associated with variation in offspring neurodevelopment, including brain
structure and connectivity, stress regulation, attention, temperament, and later emotional and
behavioural functioning (Van den Bergh et al. , 2020, 2024), and recent reviews additionally
report associations with fetal and infant brain structure and function, particularly in limbic,
prefrontal, white-matter, and connectivity-related measures, although findings remain
heterogeneous across exposures, developmental stages, and methods (Mandl et al.Where the paper says it
Where the paper says it
Where the paper says it
Developmental Integration Across Cohorts and Possible Pathways to Later Vulnerability
The Leuven cohort identified selective differences in internally organized cognitive control and its neural correlates from adolescence into adulthood <ref:2608.24983#pg2> The Tilburg cohort examined earlier sensory, affective, and attentional processing during infancy and early childhood <ref:2608. p. 7 8
Tilburg research programme developed partly from questions raised by the earlier Leuven
findings, while extending the investigation to infancy and early childhood and to more finegrained electrophysiological measures. In both cohorts, maternal anxiety was assessed prospectively during pregnancy, although the
gestational windows, instruments, and analytic approaches differed. The broader theoretical framework is developed in a companion paper (Frasch & Van den Bergh,
in preparation), which proposes that early experience across a spectrum from mild to severe
calibrates precision weighting, and considers, as separable extensions, the possible emergence of
maladaptive priors under sustained adversity and the energetic costs of maintaining high
precision. The present empirical companion addresses only the milder, non-clinical, prenatal end
of that spectrum: whether maternal anxiety during pregnancy is associated with differences in theWhere the paper says it
Where the paper says it
Interpretational Constraints and Falsifiability Considerations
A first constraint concerns evidential independence, which limits the strength of any cross-publication inference <ref:2608.24983#pg8> A second constraint concerns the specificity of PP constructs, as none of the original publications formally estimated computational precision, learning rates, uncertainty, volatility, or directional prediction-error signalling <ref:2608. p. 2 3
allocated to predictable, threat-related, ambiguous, contextual, feedback-related, and bodily
information. Because the publications did not formally estimate precision, prediction errors, or
priors, we interpret them as hypothesis-supporting evidence for a PP-informed developmental
account rather than a formal test. Specifically, we examine whether differences originally described across sensory,
affective, cognitive, neural, and interoceptive domains can be organized in terms of the
allocation of relevance and processing resources and the development of internally organized
regulation. We distinguish directly demonstrated findings from PP-informed interpretations and
identify observations that support, constrain, or could falsify this account. Specifically, we examine whether differences originally described across sensory,
affective, cognitive, neural, and interoceptive domains can be organized in terms of the
allocation of relevance and processing resources and the development of internally organized
regulation. We distinguish directly demonstrated findings from PP-informed interpretations and
identify observations that support, constrain, or could falsify this account.Where the paper says it
Where the paper says it
Where the paper says it
Conclusion
Taken together, the findings are compatible with a parsimonious PP-informed account in which maternal anxiety during pregnancy is associated with differences in how relevance and processing resources are allocated to predictable, threat-related, affectively ambiguous, contextual, feedback-related, and bodily information and in how regulation becomes internally organized <ref:2608.24983#pg8> The strongest cross-cohort conclusion is not that prenatal maternal anxiety produces a single fixed PP phenotype but rather both cohorts reveal selective developmental differences in the regulation of information relevance <ref:2608. p. 8 Aim and Research Question
We ask whether findings across eleven interrelated publications from the Leuven and Tilburg
cohorts form a selective developmental pattern associated with maternal anxiety during
pregnancy. Specifically, we examine whether differences originally described across sensory,
affective, cognitive, neural, and interoceptive domains can be organized in terms of the
allocation of relevance and processing resources and the development of internally organized
regulation. Applying this framework retrospectively is useful only if it provides more than new terminology
for previously reported associations. The potential contribution lies in examining whether
findings originally described in terms of attention, habituation, executive functioning,
connectivity, cognition, and bodily-signal processing form a selective developmental pattern
across ages and modalities.Where the paper says it
Where the paper says it
Declarations
Data availability. This article reinterprets previously published data from the Tilburg Prenatal Early Life Stress cohort and the Leuven Prenatal Project; no new data were generated <ref:2608. p. 5 However, these studies differ widely in
exposure, age, paradigm, and outcome, and few directly test whether the observed associations
reflect a common pattern of context-dependent information weighting. The present article examines this question across eleven interrelated publications from two
deeply phenotyped, prospectively followed cohorts: the Leuven Prenatal Project and the Tilburg
Prenatal Early Life Stress (PELS) cohort, hereafter referred to as the Leuven cohort and the
Tilburg cohort, respectively. 8
Tilburg research programme developed partly from questions raised by the earlier Leuven
findings, while extending the investigation to infancy and early childhood and to more finegrained electrophysiological measures. In both cohorts, maternal anxiety was assessed prospectively during pregnancy, although the
gestational windows, instruments, and analytic approaches differed.Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Precision weighting calibration for information allocation: The improved AI can allocate relevance and processing resources to sensory, affective, contextual, feedback-related, bodily information based on a precision weighting framework derived from maternal anxiety during pregnancy.
-
Internal goal maintenance and regulation modeling: The system can model how cognitive control is generated and maintained endogenously by simulating the selection of actions based on internal models of task goals and expected outcomes rather than relying solely on explicit external cues.
-
Interoceptive affect evaluation: The AI can interpret bodily signals, such as dyspnoea perception, by assigning affective meaning to them based on prenatal anxiety, allowing it to predict greater unpleasantness in these contexts without requiring generalized sensory abnormality.
-
Selective information weighting across modalities: The system can distinguish between different types of input—such as repeated standard tones versus deviant stimuli—by applying differentially weighted relevance scores, reflecting the findings that
repeated standards become increasingly expected and may therefore elicit attenuated processing as the regularity of the sequence is learned.
-
Developmental trajectory prediction: The improved AI can predict later socio-emotional and behavioral difficulties by tracking how early differences in sensory and affective weighting calibrate broader regulatory systems over time, suggesting that "Persistent processing of repetitive input, increased weighting of threat-related vocal information, or enhanced attention to affectively uncertain neutral stimuli could increase the amount of information treated as potentially relevant."
Abstract
Prenatal maternal distress has been associated with differences in offspring brain and behavioural development, but is rarely framed computationally. Predictive-processing (PP) accounts propose that perception, cognition, and action arise from interactions between input and internally generated predictions, partly governed by precision. We reappraised eleven publications from two non-clinical prospective cohorts, selected using prespecified criteria, to assess whether they form a coherent PP-compatible developmental pattern. Eight Leuven publications spanned adolescence to adulthood. Externally cued response inhibition showed no detectable group differences in modestly powered samples, whereas differences emerged when task goals, stimulus relevance, expected outcomes, and response policies had to be generated or maintained endogenously. At age 28, higher prenatal anxiety was associated with lower radial and mean diffusivity in the left superior posterior corona radiata, lower Vocabulary scores, weaker medial prefrontal-left inferior frontal and prefrontal-left supramarginal gyrus connectivity, and greater unpleasantness of induced dyspnoea. In three Tilburg ERP publications (infancy and early childhood), higher prenatal anxiety was associated with increased processing of repeated standard tones, fearful relative to happy vocalizations, and neutral rather than pleasant or unpleasant pictures. Maternal mindfulness showed the opposite association for the repeated standard. Across cohorts, findings were compatible with differences in how relevance and processing resources are allocated to predictable, threat-related, ambiguous, contextual, feedback-related, and bodily information. Because the publications did not estimate precision, prediction errors, or priors, this is a theory-guided, post-hoc reappraisal that generates discriminating hypotheses rather than a formal computational test.
Related papers
- ViroGym: Realistic Large-Scale Benchmarks for Evaluating Viral Proteins
- CellCanon: An Agentic Cellular World Model for Prediction, Explanation, and Observation-Driven Updating
- Probing and steering biology across Boltz-1s trunk-diffusion boundary
- Trajectory Inference of Human Aging from Cross-Sectional DNA Methylation Data
- Bridging the Modality Bottleneck in Pathology MIL through Molecularly Guided Feature Adaptation
- Molecular notation can act as a shortcut in AI benchmark evaluation