Protein Quality, Growth, and Cognitive Development: Why Children Need Animal-Source Nutrients for IQ and Height
A variance- and mechanism-first case that plant-only protein patterns impose predictable developmental bottlenecks unless aggressively engineered to compensate
Keywords
Child development; intelligence; IQ; height; stunting; protein quality; essential amino acids; vitamin B12; iron; iodine; zinc; DHA; vegetarian; vegan; animal-source foods; dietary patterns
Thesis
Children raised on diets that rely primarily on vegetable proteins face a higher risk of constrained cognitive development and reduced linear growth because plant-only patterns are more likely to fall short in bioavailable key nutrients and in protein quality (essential amino-acid adequacy), especially vitamin B12, iron, iodine, zinc, and preformed long-chain omega-3 fats. When these constraints occur during sensitive developmental windows, the result is a measurable downward pressure on height and on IQ-related outcomes relative to children with routine access to high-quality animal-source protein and nutrient co-factors, even when total calories appear adequate (Neumann et al., 2007; Schürmann et al., 2021; Northstone et al., 2012).
Abstract
This article develops a mechanistic and empirical case that diets centred on vegetable proteins during childhood can reduce developmental potential for both height and cognition compared with diets that provide regular, high-quality animal-source proteins. The argument is not that “plants are bad,” but that plant-only or plant-dominant protein patterns increase the probability of specific nutritional bottlenecks unless actively and consistently compensated through careful planning and supplementation. The article synthesises evidence on protein quality and limiting amino acids, nutrient co-packaging in animal-source foods (vitamin B12, iron, iodine, zinc, and long-chain omega-3 fats), and the developmental pathways linking those inputs to linear growth and neurocognitive outcomes. Observational cohort evidence shows that early-childhood dietary patterns are associated with later IQ in the population, consistent with a lasting influence of early nutrition (Northstone et al., 2012; Smithers et al., 2013). Randomised school-feeding evidence indicates that adding animal-source foods can improve cognitive performance and behavioural outcomes in settings where diets are limited (Neumann et al., 2007). The article then sets out an analysis plan for comparing vegetarian, vegan, and omnivorous child diets using standardised IQ measures and height-for-age outcomes, with explicit controls for socioeconomic confounding, parental education, and overall diet quality. The conclusion states testable predictions: plant-only patterns will show a higher prevalence of nutrient inadequacy markers, lower mean height-for-age, and small but detectable IQ penalties when supplementation and nutrient adequacy are not reliably maintained (Schürmann et al., 2021; OUP scoping review, 2025).
1 — The claim stated cleanly: IQ and height are nutritional outcomes
This article advances a precise, testable claim about childhood development. The comparison is between children raised on vegetarian or vegan dietary patterns and children who have routine access to high-quality animal-source proteins, and the outcomes are not vague impressions of “health” but two measurable developmental endpoints: height, captured through standardised growth metrics (such as height-for-age measures), and cognitive performance, captured through standardised IQ instruments and closely related cognitive composites. The claim is not moral, cultural, or aesthetic. It is a developmental constraint claim: the inputs available during growth and neurodevelopment impose limits or affordances that can be measured later as differences in linear growth and in cognitive performance.
Figure 1. Conceptual pathway map: protein quality and micronutrients → growth and neurodevelopment → height and IQ-
This figure illustrates the conceptual outcome of the developmental pathway described in Section 1. Two IQ distributions are shown, representing children with routine access to high-quality animal-source proteins and children raised on an unplanned vegetarian diet.
-
The meat-access distribution is centred at a mean IQ of 101 with a standard deviation of 15, reflecting adequate protein quality and micronutrient availability during development. The unplanned vegetarian distribution is shifted left, with a mean of 96 and a slightly wider spread (SD = 16), representing a higher probability of nutritional bottlenecks affecting neurodevelopment. The overlap between the curves emphasises that many children in both groups function within the normal range; the shift in the centre and the spread reflect population-level effects rather than deterministic outcomes for individuals.
-
The figure is not a claim about ideology or inevitability. It operationalises the article’s central thesis: differences in protein quality and micronutrient reliability during development act upstream of growth and brain development, and those upstream differences can translate into measurable shifts in height and IQ distributions at the population level.
The argument begins by refusing a common simplification: protein is not merely “grams per day.” A child can meet an arbitrary gram target and still fail to meet the biological requirements that matter for development. What matters is protein quality: whether essential amino acids are available in adequate proportions, whether digestion and absorption make those amino acids bioavailable, and whether the diet reliably supplies the co-factors that allow tissue growth and neurodevelopment to proceed without bottlenecks. When diets are built around animal-source foods, these constraints are typically met with less engineering because animal-source proteins tend to deliver complete amino-acid profiles and high digestibility while also co-packaging several micronutrients that are repeatedly implicated in child development. When diets exclude animal-source foods, the constraints do not disappear; they become a logistics problem that must be actively solved every day.
For the purpose of this article, “high-quality animal-source proteins” does not mean excess meat consumption or a particular ideology. It means routine access to food sources that provide dense, bioavailable protein and the micronutrients that commonly ride with those proteins in real diets: vitamin B12, iron in more readily absorbed forms, zinc, iodine-related adequacy through diet patterns, and preformed long-chain omega-3 fats such as DHA and EPA. These nutrients are not optional decorations. They are repeatedly discussed in the paediatric nutrition literature because shortfalls during development have plausible, and in some cases well-documented, consequences for brain development, growth, and function. The point is not that plants cannot contribute to a healthy diet; the point is that when a child’s protein pattern is predominantly plant-based, the diet must be deliberately engineered to avoid falling into predictable deficiency traps.
That leads to the article’s central risk statement. Vegetarian and vegan patterns are not automatically deficient, but they are structurally more fragile in children because they require higher precision to maintain nutrient adequacy. A well-planned plant-only diet can be made developmentally adequate, but “well-planned” is doing all the work. It requires consistency, supplementation (especially vitamin B12 in vegan patterns), careful attention to iron and zinc bioavailability, adequate iodine intake, and deliberate sourcing of long-chain omega-3 or reliable alternatives. The risk is that, in real households, precision is not constant across time, caregivers, or contexts. Diets that are “vegetarian” or “vegan” in identity can drift into a pattern that looks sufficient by calories and by crude protein grams while still falling short in amino-acid balance, bioavailability, and micronutrients that matter for neurodevelopment.
The claim of this paper is therefore not “vegetables make children stupid.” It is sharper and more difficult for critics to dodge: when children are raised on diets where protein is predominantly vegetable-derived and animal-source inputs are absent or minimal, the probability of biologically meaningful nutritional bottlenecks rises unless the diet is deliberately engineered and supplemented. Those bottlenecks are predictable rather than mysterious, and they are positioned exactly where development is most sensitive: linear growth and cognitive development. Growth becomes an external marker of adequacy because height is a visible outcome of sustained nutritional sufficiency across years; cognition becomes the deeper marker because the brain is a high-demand system built under constraints. If a diet routinely falls short of what development requires, it should not be surprising to see signals in both height and IQ.
The point of opening this way is to keep the discussion on measurable outcomes and mechanistic plausibility rather than on social signalling. A child does not benefit from dietary ideology. A child benefits from adequate building blocks, delivered reliably through development. When animal-source proteins are part of the routine diet, adequacy is easier to achieve by default. When the diet is plant-only, adequacy is possible but conditional. That conditionality is the core of the article’s thesis: plant-only patterns can be viable, but only when deliberately engineered; otherwise they raise the odds of predictable deficiencies with relevance to growth and cognition, and those deficiencies, when sustained during development, can translate into measurable differences in height and IQ (Schürmann et al., 2021; Lemale et al., 2023).
2 — What the outcomes look like: height curves, IQ distributions, and practical effect sizes
This article treats height and IQ as developmental outcomes that can be measured cleanly and reported without mysticism. The core question is not whether a dietary label sounds virtuous or fashionable, but whether diet reliably delivers the biological inputs needed for growth and neurodevelopment. For that purpose, height and IQ are ideal because they are standardised, widely used, and interpretable. They also allow the article to report effects in units that mean something outside academic theatre: centimetres of growth, z-scores relative to age norms, IQ points, and proportions of children falling below functionally important thresholds.
Height outcomes are measured using three linked metrics. The first is height-for-age z-score (HAZ), which expresses a child’s height relative to an age- and sex-referenced population standard. A HAZ of 0 is average for the reference; negative values indicate shorter stature relative to age peers. The second metric is stunting thresholds, typically defined operationally as HAZ below a predetermined cut-off that marks clinically meaningful growth faltering. In practical terms, this is the point where “a little shorter” becomes “developmentally constrained.” The third metric is growth velocity, which looks at change over time rather than a single measurement. Growth velocity matters because two children can have the same current height but different trajectories: one steadily tracking an expected growth curve and another drifting downward over successive measurements. If diet is constraining development, growth velocity is often where the constraint becomes visible before it hardens into a permanent height gap.
Figure 2. Conceptual comparison of growth and cognitive outcomes by diet quality-
This schematic presents two paired distributional comparisons to illustrate how dietary adequacy can manifest across developmental outcomes.
-
The left panel shows height-for-age z-score distributions. Children with routine access to high-quality animal-source proteins are represented by a distribution centred closer to the reference mean, with a tighter spread. Children raised on an unplanned vegetarian diet are represented by a distribution shifted leftward, indicating lower average height-for-age and a higher probability of falling into growth-faltering ranges. The overlap emphasises that many children in both groups grow within normal bounds; the shift reflects population-level risk rather than deterministic outcomes.
-
The right panel shows IQ distributions using the same conceptual grouping. The meat-access distribution is centred at a higher mean with a slightly tighter spread, while the unplanned vegetarian distribution is shifted left with marginally greater dispersion. As with height, the key feature is not categorical separation but a shift in centre and tails: small mean differences translate into meaningful differences in the proportion of children below educationally relevant thresholds and in the upper tail.
-
Together, the two panels make the article’s methodological point explicit. Height and IQ are treated as parallel developmental readouts. When diet quality is reliably sufficient, both distributions cluster closer to the normative centre. When diet quality is fragile or inconsistently adequate, both growth and cognition show leftward pressure and altered tail structure, consistent with shared underlying nutritional constraints rather than lifestyle identity.
IQ outcomes are measured using standardised instruments and composites designed to capture general cognitive performance. The article will use IQ as an umbrella label for standardised cognitive scores derived from well-established instruments such as Wechsler-style composites (e.g., full-scale IQ) or validated cognitive batteries that produce normed composite scores. The operational requirement is standardisation: scores must be age-normed and comparable within age bands. The article will not rely on informal school grades or subjective impressions. It will report cognitive outcomes using the same logic applied to height: not merely “average score,” but distributional shape and tail outcomes.
This matters because small average differences can produce large social and educational consequences once distributional consequences are made explicit. A modest mean shift—say, five IQ points—can look trivial to readers who treat IQ as a vanity statistic. It is not trivial when the issue is eligibility thresholds, special educational needs identification, high-achievement programmes, and the frequency of children who fall into the very low or very high ends of the distribution. The article will therefore report both mean differences and distributional consequences. Means answer whether one group is shifted relative to another; distributions answer how many children cluster near the centre, how many fall below functional thresholds, and how many appear in the upper tail.
Table 1. Outcome definitions and reporting metrics -
Notes for use in the article: thresholds (for stunting and IQ cut-offs) must be defined once, up front, and then held constant across cohorts and age bands; all outputs are reported by age band to avoid mixing distinct developmental distributions.
The reporting frame is deliberately pragmatic. For IQ, effects will be reported in IQ points and then translated into the proportion of children below thresholds that matter in practice. These thresholds will be defined in advance, rather than selected opportunistically. A low threshold can represent the zone where learning support becomes statistically more common; a high threshold can represent the zone used by selective academic tracks. The point is not to pretend that one number captures a child’s value. The point is to show how nutrition-related shifts change the probability of developmental difficulty and the probability of exceptional outcomes at the population level. Alongside this, the article will report standard deviations and tail proportions, because dispersion can amplify or blunt the practical impact of a mean shift. A five-point drop with unchanged variance yields one pattern; a five-point drop with increased variance yields a different one, particularly at the low end.
For height, effects will be reported as differences in centimetres and as differences in HAZ. The centimetre difference is the intuitive anchor; the z-score difference is the standardised anchor that allows comparison across ages and populations. Stunting prevalence will be reported because stunting is a binary event of developmental concern, and changes in stunting prevalence convey the real-world meaning of small shifts in average height. Growth velocity will be reported where longitudinal data exist, because it distinguishes short-but-healthy trajectories from trajectories that reflect ongoing constraint.
The reason for pairing height and IQ is methodological, not rhetorical. Height provides an external readout of whether a diet is consistently adequate over time. A diet that is claimed to be “adequate” but is associated with lower growth velocity or lower HAZ is signalling a constraint that is unlikely to spare neurodevelopment. Conversely, a diet that maintains robust growth makes it harder to argue that the child is nutritionally constrained in a way that would systematically depress cognition. Height therefore functions as a validation channel: it helps identify whether the diet is merely different in ideology or different in biological adequacy.
This section also places the article’s claims in the context of existing empirical work showing that early diet quality and nutrition correlate with later cognitive outcomes. Large cohort studies have reported associations between childhood dietary patterns and later IQ, supporting the general proposition that early nutrition is not a cosmetic variable but a measurable influence on later cognitive performance (Northstone et al., 2012). Work examining dietary trajectories from infancy through toddlerhood similarly reports that early dietary patterns have associations with later cognitive outcomes, reinforcing the idea that early-life diet can track into later developmental performance rather than washing out completely (Smithers et al., 2013). These findings do not, by themselves, prove that vegetarian diets lower IQ, because cohort studies are vulnerable to confounding. They do, however, establish a crucial baseline: diet quality in early life is empirically linked to later cognitive outcomes in real populations, which makes a mechanistic argument about protein quality and micronutrient reliability both plausible and worth testing rather than dismissing as speculation.
The rest of the article will treat these outcomes with the discipline they require. It will define groups carefully, measure adequacy rather than merely identity, and report both average differences and distributional consequences. The aim is not to posture. The aim is to show, in interpretable metrics, what height and IQ outcomes look like when childhood diets reliably meet developmental requirements versus when they rely on plant proteins and micronutrient logistics that are easier to get wrong, particularly when a diet is “unplanned” rather than engineered.
3 — Mechanism I: protein quality is not optional biology
The slogan “protein is protein” works as social comfort, not as developmental biology. In child development, the relevant question is not whether a child consumed a number that looks like “enough grams,” but whether the child received the amino acids required to build and maintain rapidly expanding tissues, including the brain, under conditions of continuous growth. Protein is not a single substance. It is a delivery vehicle for amino acids, and amino acids are the units the body actually needs. If any essential amino acid is insufficient relative to demand, the body cannot simply substitute another. It cannot “work around” a missing essential amino acid any more than a builder can replace bricks with paint. The shortage becomes a bottleneck. That is why protein quality matters, and why it matters more in children than in adults.
Figure 3. Amino-acid adequacy funnel: total grams → digestibility → limiting amino acids → effective protein for growth/brain-
This schematic shows why counting “protein grams” is an incomplete developmental metric. The funnel represents successive constraints that reduce paper protein into effective protein available for child growth and brain development.
-
At the top is total protein grams, which is what diet labels and casual tracking usually emphasise. The next narrowing is digestibility and absorption, because not all consumed protein becomes usable amino acids. The next narrowing is the critical bottleneck: limiting essential amino acids (conceptual anchors such as lysine and methionine). If one essential amino acid is relatively scarce, it constrains the utilisation of the rest, reducing the effective amino-acid pool available for synthesis. At the bottom is effective protein for growth and brain, the fraction that actually supports sustained linear growth and neurodevelopment.
-
The side callouts state the developmental implication: child demand is continuous, and growth and neurodevelopment draw from the same amino-acid pool. The practical conclusion is direct: a diet can appear “high protein” while still failing an essential-amino-acid adequacy test if digestibility and limiting amino acids are not secured.
Start with the essential amino acids. These are amino acids the body cannot synthesise in adequate quantities and therefore must obtain from diet. For a growing child, essential amino acids are required continuously because growth is continuous. There is no pause button. Muscle, bone matrix, connective tissue, blood proteins, enzymes, and the structural and signalling proteins of the developing nervous system all draw from the same amino acid pool. A child’s body is not optimising for “maintenance.” It is building. That is why a child can appear “well-fed” by calorie intake and still show constraint: calories can keep weight up, but calories cannot create missing amino acids.
The second step is to understand why limiting amino acids matter more than total protein grams. A dietary protein is “complete” in practical terms only if it supplies essential amino acids in adequate proportions relative to the body’s needs. If one amino acid is low, it limits the utilisation of others. This is the bottleneck logic: synthesis of many proteins stalls when one required component is missing or scarce. In plant-forward patterns, the limiting amino acids are not random. Certain plant protein sources are commonly lower in lysine, while others can be relatively lower in methionine or have imbalances that require careful combination. The precise pattern varies across foods, but the conceptual point holds: plant proteins often require pairing, quantity, and consistency to reach the same effective amino-acid adequacy that animal-source proteins tend to provide more reliably by default. A vegetarian diet can meet amino-acid needs, but it must be constructed to do so. An unplanned vegetarian pattern increases the probability that the child’s “protein grams” conceal an amino-acid gap.
Table 2. Mechanistic checklist: requirements for a vegetarian/vegan child diet to be developmentally equivalent-
Use rule for the article: a vegetarian/vegan diet is classified as “developmentally equivalent” only when the checklist is demonstrably satisfied consistently over time; otherwise it is classified as “unplanned/high-risk,” which is the group the thesis predicts will show the greatest height and IQ penalties.
Lysine is a useful conceptual anchor because it frequently emerges as a limiting amino acid in diets heavily reliant on certain cereal grains. If lysine intake is marginal, growth-related protein synthesis is constrained even if total protein intake looks superficially adequate. Methionine provides a second anchor because it often requires deliberate attention in plant-only patterns, and because methionine and related pathways feed into broader metabolic processes beyond simple structural protein synthesis. The intent here is not to turn an article into a biochemistry lecture. It is to make the developmental logic unavoidable: when essential amino acids are marginal, the body prioritises survival and core function over growth and over the optimisation of higher-order development. That trade-off is not ideological. It is forced by constraints.
A third step is digestibility, because protein quality is not only about amino-acid composition; it is also about how much of that protein is actually absorbed and usable. In practical terms, two diets can contain identical “protein grams” on paper but deliver different amounts of usable amino acids to the body. Animal-source proteins are generally highly digestible and deliver amino acids in forms that are readily absorbed. Many plant proteins are also digestible, but plant-forward patterns can include factors that reduce digestibility and absorption, especially when the diet is built around staples and ultra-repetitive food choices. This is not a claim that plants are poison. It is a claim that the effective protein delivered to a growing child is the product of composition and absorption. When either is compromised, the deficit is not corrected by “counting grams.” It is corrected only by changing the diet’s amino-acid adequacy and bioavailability.
Now connect this to child development directly. Growth and brain development are concurrent high-demand processes. Linear growth is visible and measurable, and it is metabolically expensive. Bone growth requires collagen and other structural proteins, muscle growth requires continuous synthesis, and the endocrine environment that regulates growth is sensitive to nutritional adequacy. At the same time, the brain is undergoing rapid development: synaptic formation, myelination, neurotransmitter system maturation, and the construction of structural and functional networks that will support later cognition. These processes are not optional extras. They are the core of development. When amino-acid supply is marginal, the body faces an allocation problem: the same pool must serve growth, immune function, organ maintenance, and neural development. Something gives first, and when constraints persist, they become visible in outcomes.
Height is therefore not treated as a vanity metric in this article. Height is treated as a canary. If a child’s diet is limiting in usable protein quality, linear growth is one of the most likely systems to show constraint over time. A persistent shortfall in amino-acid adequacy does not always cause dramatic failure, but it can cause a subtle downward pressure on growth velocity. Over months and years, small downward pressures accumulate. The outcome can be a lower height-for-age trajectory even when calories are adequate and even when weight looks acceptable. That matters because the same dietary pattern that constrains growth is unlikely to spare the brain. The logic is not that “short children are less intelligent.” The logic is that both height and cognition are downstream of the same adequacy constraints, and height provides an external marker that those constraints are real.
The brain-specific pathway can be described without melodrama. Neurotransmitters are synthesised from amino acid precursors, and the proteins involved in neural structure and function must be synthesised continuously during development. The brain does not only need energy. It needs materials. And because the brain’s development is time-sensitive, shortages during sensitive windows can have lasting consequences. This is one reason the article insists on child development rather than adult nutrition. Adults can compensate with stability, redundancy, and reserves; children are building the system itself. If the system is built under constraint, later compensation is limited. A protein gap in a growing child is therefore not comparable to a protein gap in a stable adult.
This also explains why “but they ate enough calories” is an inadequate rebuttal. Calories prevent starvation. They do not guarantee development. The child development question is whether the diet supplies not only energy but the correct building blocks. A child can eat large volumes of low-quality protein sources, meet caloric targets, and still run into amino-acid bottlenecks. A child can also meet protein grams in a plant-forward diet and still run into bottlenecks if the diet is narrow, repetitive, or inconsistently planned, because adequacy is not achieved by slogans; it is achieved by precision.
At this point the mechanistic claim becomes sharp enough to connect to the article’s empirical frame. The thesis predicts that “unplanned vegetarian” patterns—meaning diets that exclude animal-source proteins without consistent, deliberate optimisation of amino-acid completeness and bioavailability—will show signs of constraint more often than diets with routine access to high-quality animal-source proteins. Those signs will appear first as higher rates of micronutrient inadequacy and, in many cases, as subtle reductions in growth velocity and height-for-age. The cognitive outcomes are expected to track the same constraint pattern, not because diet labels magically determine intelligence, but because development is limited by inputs. When the inputs are consistently adequate, plant-based diets can work. When the inputs are fragile, the child pays the price, and the price is measurable.
The purpose of this section is to strip the discussion down to biology and logistics. Child development is not a debate club. It is a system under construction. Protein quality is not a philosophical preference. It is a constraint boundary. If that boundary is crossed—if essential amino acids and usable protein supply are marginal—then growth and neurodevelopment are forced to operate in a restricted mode. That restricted mode need not produce catastrophic deficits to matter. In population terms, it produces a shift: slightly lower growth trajectories and slightly lower cognitive outcomes, with real consequences once those shifts are mapped onto thresholds and distributions.
4 — Mechanism II: the micronutrient co-packaging problem (B12, iron, iodine, zinc, DHA)
The decisive difference between plant-only child diets and diets that include animal-source foods is not ideology. It is logistics. The child’s body does not care why a nutrient is absent; it only registers whether the nutrient is reliably present during development. Animal-source foods matter in this argument because they solve multiple logistics problems at once: they tend to deliver high-quality protein and a cluster of developmentally relevant micronutrients in bioavailable forms, without requiring a daily engineering mindset. Plant-only patterns can be made adequate, but adequacy becomes a continuous project, and the project has known failure points. In child development, those failure points are not trivial because small deficits sustained across time can translate into measurable constraints in growth and neurocognitive function.
The central concept is “co-packaging.” Many of the nutrients most repeatedly implicated in child growth and brain development are naturally present in animal-source foods at densities and in forms that are comparatively straightforward to use. When those foods are removed, the nutrients do not disappear from biology; they simply become harder to secure reliably. Families must replace the co-packaged nutrient bundle with deliberate combinations of plant foods, fortified foods, and supplements, and they must do so continuously, through changing appetites, illness periods, school routines, travel, and caregiver variability. That is why unplanned vegetarian or vegan patterns are high-risk: not because plant foods are intrinsically harmful, but because the burden of execution is higher, and the consequences of execution failure fall on the child’s development.
Vitamin B12 is the clearest example because it is the least negotiable. B12 is not a “nice to have.” It is a cornerstone nutrient for nervous system development and function, and it is structurally difficult to obtain from unfortified plant sources in reliable amounts. For vegan children, supplementation or consistent use of fortified foods is not optional. If it is omitted, the diet is not “a vegan diet with a different philosophy”; it is a nutritionally incomplete diet. The importance of B12 is not a matter of online debate. It is a practical constraint: B12 must be present and must remain present through development. This is why supplementation adherence is the hinge variable in vegan child nutrition: a vegan pattern with reliable B12 provision is one thing; a vegan pattern without it is a predictable deficiency trap. Reviews focusing on vegan children repeatedly identify B12 as a central vulnerability and treat supplementation as the essential mitigation, precisely because the diet itself does not solve the B12 logistics problem without fortification or supplementation (Schürmann et al., 2021).
Iron illustrates a second class of logistics problem: not only intake, but bioavailability. Iron is developmentally relevant because the growing child is expanding blood volume, building tissues, and supporting brain development. The child’s iron needs are not a static adult maintenance requirement. Iron also varies in how readily it is absorbed, and diet patterns can either facilitate or inhibit absorption. Animal-source foods often provide iron in forms that are easier to absorb, while plant-only patterns can deliver iron but frequently face absorption constraints unless meals are deliberately structured. The argument here is not that plant iron “does nothing,” but that a plant-only diet increases the probability of marginal status if the diet is repetitive, energy-adequate but micronutrient-thin, or dominated by foods that do not support absorption effectively. That probability increase is the relevant point because development is sensitive to marginality sustained over time. A child whose iron status repeatedly dips during growth is operating in a constrained mode. In real-world terms, this is exactly the kind of pattern that can coexist with outward normality until it accumulates into measurable differences in growth and cognitive performance.
Figure 4. “Co-packaged nutrients” versus engineered adequacy in child diets-
This diagram contrasts two fundamentally different nutritional logistics models.
-
On the left, diets that include routine animal-source foods deliver co-packaged nutrients by default: complete, highly digestible protein alongside vitamin B12, iron in more bioavailable forms, zinc, typical iodine adequacy through common food patterns, and preformed long-chain omega-3 fats (DHA/EPA). Adequacy emerges as a side effect of ordinary eating rather than as a continuous optimisation task.
-
On the right, a plant-only pattern can reach developmental adequacy only through deliberate engineering. Protein quality must be actively managed through pairing and density; vitamin B12 must be supplied via supplementation or fortification; iron and zinc require absorption-aware strategies; iodine must be tracked via iodised salt or supplements; and DHA/EPA must be sourced deliberately through supplements or fortified foods. The arrow between the two panels represents increased logistics burden and compliance sensitivity.
-
The figure makes the article’s central point explicit: the biological difference is not ideology but execution reliability. When adequacy depends on continuous planning and adherence, developmental risk rises whenever that execution falters.
Iodine is often neglected in casual discussions because it is not visible until it becomes a problem. In development, iodine matters because it supports thyroid hormone production, and thyroid hormones sit upstream of both growth and brain development. This is not an abstract claim; it is a pathway claim: iodine adequacy supports thyroid function, thyroid function supports growth and neurodevelopment, and inadequate iodine creates a risk pathway that cannot be repaired by calories alone. When diets include animal-source foods and typical fortified patterns, iodine adequacy is easier to achieve as a side effect. When diets exclude common iodine sources, iodine can become an inadvertent blind spot, especially if households do not consistently use iodised salt or do not track iodine sources deliberately. The logistic point is again central: plant-only diets can meet iodine needs, but they do not automatically meet them, and a diet that appears “healthy” in other respects can still fail on iodine if the household does not treat iodine as a deliberate target.
Zinc belongs in the same category as iron from a practical standpoint: it supports growth and development, but plant-forward patterns can make adequacy more difficult to secure consistently if diets are narrow and dominated by staple foods without intentional zinc density. Zinc is not merely “immune function” in a hand-waving sense; it is a growth-linked nutrient, and childhood growth is exactly where zinc adequacy matters most. In an unplanned vegan pattern, zinc can be marginal without obvious daily symptoms, and the child’s trajectory is where the effect is observed: slower growth velocity, lower height-for-age, and broader vulnerability during illness periods when intake falls. This again is why the article treats height as a companion outcome: growth is where micronutrient constraints become visible over time.
Table 3. Nutrient risk matrix (child development): deficiency risk, developmental relevance, mitigation route, compliance sensitivity-
Interpretive rule for the article: the thesis predicts the largest IQ and height penalties in the subgroup where multiple “high compliance sensitivity” items fail simultaneously (especially B12 and DHA/EPA, with iron/iodine as common co-failures), rather than uniformly across all vegetarian households.
Long-chain omega-3 fats, especially DHA and EPA, illustrate the co-packaging problem in its purest form. The brain is lipid-rich, and neural membranes rely on fatty acid composition. A plant-only diet can include short-chain omega-3 sources, but the practical question for development is not whether an adult can survive without dietary DHA/EPA; it is whether developing brains are best supported when preformed long-chain omega-3 sources are reliably present. Diets with animal-source foods can supply DHA/EPA more readily, while vegan diets require deliberate sourcing through supplements or fortified products to secure preformed long-chain omega-3 intake reliably. The issue again is not theoretical possibility. It is execution reliability. If a diet requires an engineered solution to deliver DHA/EPA consistently, then adherence and household competence become developmental variables. That is not a comfortable conclusion for lifestyle marketing, but it is the relevant conclusion for child outcomes.
This is why the article treats supplementation adherence as the hinge variable rather than treating diet identity as destiny. A vegan child diet that is engineered—meaning adequate energy, adequate high-quality protein via deliberate food combinations, consistent B12, deliberate attention to iron and zinc bioavailability, reliable iodine, and reliable DHA/EPA provision—can be developmentally adequate. A vegan child diet that is not engineered becomes a predictable deficit risk pattern. The empirical literature on vegan children, including work focusing on nutrient status and growth, repeatedly centres this point: the key concern is not “veganism” as a label but the elevated risk of deficiencies when the diet is poorly planned or when supplementation is inconsistent (Schürmann et al., 2021). Broad reviews of plant-based diets in children in high-income settings make the same directional point while also emphasising heterogeneity and the need for better causal evidence: outcomes depend on planning quality, food composition, and supplementation, and the current evidence base does not permit simplistic generalisations (Petersen et al., 2024; OUP scoping review, 2025).
The logic of co-packaging therefore becomes a straightforward prediction. Diet patterns that include routine animal-source foods should show fewer adequacy failures because the nutrients are delivered by default. Plant-only patterns should show a wider range of outcomes, with the lowest outcomes clustering where supplementation is inconsistent and where diet quality is narrow or ultra-processed. That is the mechanism this article will test. It does not depend on demonising plants. It depends on a simple proposition: child development is time-sensitive, and nutrients that are hard to secure reliably in plant-only patterns become the predictable points at which neurodevelopment and growth are constrained when the diet is “unplanned” rather than engineered.
5 — What the empirical literature actually shows about diet and IQ
The evidence linking diet to cognitive development does not come in one neat package. It arrives in three distinct classes, each with different strengths and failure modes. The correct way to build a serious argument is to separate them, state what each can and cannot support, and then extract a convergent logic that is strong enough to test. The thesis advanced in this article does not require pretending the literature is cleaner than it is. It requires showing that the best available evidence is consistent with a mechanistic claim about nutrient adequacy and protein quality, and that the sharpest empirical tests can be specified rather than improvised.
The first evidence class is cohort evidence. In UK birth-cohort work, early-childhood dietary patterns have been associated with later IQ outcomes. Northstone et al. (2012) reported that childhood dietary patterns at early ages were associated with IQ at age 8, with less favourable patterns tracking with lower later scores. The critical value of such work is not that it “proves” any single causal pathway; the value is that it establishes an empirical baseline: early diet is not merely a cosmetic preference variable. It correlates with later cognitive performance in a way that is measurable, persistent, and aligned with a general developmental intuition that early inputs matter. Smithers et al. (2013) reinforce this baseline through a related approach, examining dietary trajectories across infancy and toddlerhood and reporting associations with later IQ in childhood and adolescence. Taken together, these cohort results support a central proposition required for the thesis: nutrition in early life has lasting cognitive correlates that do not vanish into noise.
Cohort evidence, however, comes with a predictable weakness: confounding. Families do not select diets randomly, and the same social factors that influence dietary choices can influence cognitive outcomes through multiple channels. Parental education, socioeconomic status, parental cognitive ability, home learning environment, birth and prenatal factors, and broader health behaviours cluster together. In affluent contexts, vegetarian households can be more educated and health-conscious, which can bias naïve comparisons in a direction opposite to the thesis. That means cohort data do not settle the argument by themselves. What they do is justify treating diet as a plausible contributor and justify building designs that can separate “diet identity” from “diet adequacy.” In other words, cohort evidence supports the problem statement—early diet matters—while also forcing discipline: any claim about vegetarian diets and IQ must be tested in a way that survives the confounding structure of real families.
The second evidence class is intervention evidence, and it matters because it shows plausibility under more controlled conditions. Neumann et al. (2007) examined school-feeding supplementation with animal-source foods in Kenyan children and reported improvements in outcomes that included cognitive and behavioural measures. The point here is not to claim that a Kenyan school-feeding trial directly maps onto an affluent child in a high-income setting. The point is to demonstrate that changing diet inputs—specifically adding animal-source foods—can move functional cognitive outcomes in a real-world child population. This strengthens the causal plausibility of the general mechanism: when baseline diets are constrained, improving protein quality and nutrient delivery can improve developmental performance. It also shows something important about the thesis’s structure: the effect is expected to be strongest where diets are marginal. That is precisely the thesis’s prediction for “unplanned vegetarian” patterns. The thesis is not that every vegetarian child is harmed. The thesis is that children are harmed when the diet’s adequacy is fragile, and interventions show that improving nutrient delivery can improve outcomes in constrained environments.
The intervention class also has limitations that must be stated cleanly. Trials are often context-specific, they may measure academic performance rather than full-scale IQ, and the causal agent is frequently a bundle: protein quality plus micronutrients plus energy density plus attendance effects. This is not a defect; it is reality. But it does mean intervention evidence is best used to support a mechanism and a directional plausibility claim, not to provide a definitive effect size for IQ in a different context. It supplies the “can diet move cognitive outcomes?” answer, which the thesis requires.
The third evidence class is the direct vegetarian/vegan child literature, and this is where honesty is most crucial. Reviews focusing on vegan children repeatedly show two things at once: vegan diets can support healthy growth when carefully planned, and deficiencies are a central risk when diets are poorly planned or supplementation is inconsistent. Schürmann et al. (2021) emphasise nutrient status and growth in vegan children, focusing on the predictable vulnerability points and the practical reality that supplementation adherence is the hinge variable. Lemale et al. (2023) similarly address diet and growth in vegetarian and vegan children, emphasising that outcomes depend on diet quality and planning rather than on the label alone. A broader scoping review in Nutrition Reviews (2025) frames the same direction: paediatric outcomes under vegan patterns vary with planning quality, fortification, supplementation, and overall diet structure.
This literature is not a gift to either side of the argument. It does not support a crude claim that vegetarianism automatically reduces IQ. It does support the article’s central framing: plant-only child diets can be adequate, but adequacy is conditional, and the risk profile concentrates exactly where the thesis predicts it should concentrate—where planning is inadequate and supplementation fails. That matters because the thesis is not a metaphysical claim about “plants.” It is a logistics claim about the probability of adequacy failures across real households. A diet that requires precision is, in practice, a diet with a larger variance in execution quality. That variance in execution quality is expected to produce variance in child developmental outcomes, and the lower tail of execution quality is expected to carry measurable penalties in both growth and cognition.
Figure 5. Evidence map: cohort associations vs interventions vs diet-comparison studies (annotated strengths and confounds)-
This figure organises the evidence base into three classes and makes explicit what each class can contribute to the argument, and what each class cannot do without further controls.
-
Cohort evidence provides scale, real-world diet exposure, and longitudinal follow-up, but is vulnerable to confounding from socioeconomic status, parental cognitive ability, and home environment (Northstone et al., 2012; Smithers et al., 2013). Intervention evidence provides causal leverage because the exposure is manipulated, but it is often context-specific and may bundle multiple nutrients rather than isolating “protein source” alone (Neumann et al., 2007). Diet-comparison studies that directly examine vegetarian/vegan children are the most directly relevant, but they frequently involve heterogeneous planning quality and supplementation adherence, with small or selective samples that can blur effects (Schürmann et al., 2021; Lemale et al., 2023; Influence of a vegan diet on child health and development: A scoping review, 2025).
-
The convergent logic is therefore not a slogan. It is a methodological structure: cohorts establish association; interventions establish plausibility; diet-comparison studies locate the hinge variable—nutrient adequacy and planning compliance—where the thesis predicts the IQ and height penalties should cluster.
Put together, the three evidence classes converge on a coherent, testable statement. Cohort studies show that early diet quality correlates with later IQ, indicating that nutritional inputs track into cognitive outcomes (Northstone et al., 2012; Smithers et al., 2013). Intervention work shows that adding animal-source foods can improve functional cognitive outcomes in diet-limited settings, supporting causal plausibility for nutrient delivery affecting development (Neumann et al., 2007). Vegan/vegetarian paediatric reviews show that outcomes hinge on planning and supplementation, and that deficiencies are the central risk in poorly planned diets (Schürmann et al., 2021; Lemale et al., 2023; Nutrition Reviews, 2025). The thesis does not require pretending the literature already contains a perfect head-to-head IQ experiment in affluent vegetarian families. It requires making the correct prediction: the IQ penalty, if present, will track nutrient adequacy and protein quality, not self-identification.
Stated in testable form: children raised on unplanned vegetarian or vegan patterns—defined operationally by documented shortfalls in B12 provision, iron or zinc status risk, iodine inadequacy risk, and absent or inconsistent preformed DHA/EPA provision, alongside weaker protein-quality proxies—will show lower average IQ scores and higher proportions below educationally meaningful thresholds than children with routine access to high-quality animal-source proteins, even after adjustment for socioeconomic and parental factors; and this difference will attenuate or vanish in the subgroup of vegetarian/vegan children whose diets meet adequacy criteria consistently over time.
6 — Height as the external validator: why growth is the canary
Height is not a vanity metric in this article. It is a developmental readout. Linear growth is one of the clearest external indicators of whether the body has been receiving the materials and co-factors required for sustained development across time. That is why height-for-age measures are used globally in child health and development work: they are not a judgement of worth; they are a record of whether a child’s environment—including diet—has supported growth without persistent constraint. When a diet is limiting in usable protein quality or in critical micronutrients, linear growth is often among the first systems to show the consequences, not necessarily as dramatic failure, but as a subtle drift in growth velocity that becomes visible only when measurements are repeated.
Figure 6. Joint-outcome plot concept: height-for-age z-score vs IQ, with diet groups overlaid-
This figure shows the conceptual joint-outcome structure the article reportsAddendum Prologue — Terms, execution realities, and the economics of “unplanned”
In this article, “unplanned” is a technical term about execution, not a claim about intent. It does not mean accidental, careless, or unaware. It means the diet is not run as an engineered system with explicit targets, verified inputs, and routine monitoring. A household can be intensely committed to veganism and still be “unplanned” in the relevant sense if it does not consistently secure the nutrients that vegan child development requires. Conversely, a household can be indifferent to dietary ideology and still be “planned” if it reliably hits the biological constraints through deliberate food choice, fortification, supplementation, and follow-through.
The article therefore uses a set of distinct terms to avoid confusion. “Diet identity” refers to what foods are excluded or included: omnivorous, vegetarian (lacto-ovo), and vegan (plant-only). “Diet adequacy” refers to whether the pattern meets developmentally required inputs over time. “Engineered adequate” refers to diets that meet adequacy criteria consistently—adequate energy density, adequate effective protein (amino-acid completeness plus digestibility), and deliberate management of the micronutrients that do not reliably appear in plant-only patterns. “Unplanned/high-risk” refers to diets that are not managed with that engineering discipline, whether because the household lacks information, time, money, continuity of caregivers, or because the household believes that “whole foods” automatically solve the logistics. The term is chosen because the risk is structural: when adequacy requires repeated precision, a diet without systems is statistically more likely to drift into shortfall.
This is why the article repeatedly separates “unplanned” from “unintentional.” Many vegan households intend the best possible outcome. The problem is that intent does not generate B12, iodine, or DHA/EPA. Implementation does. An unplanned vegan pattern is often characterised by one or more of the following: inconsistent or absent B12 supplementation; reliance on food choices that are nutritionally virtuous but not micronutrient-dense; avoidance of fortified products because they are perceived as “processed”; lack of an iron or zinc strategy; absence of iodised salt as a stable iodine source; and no deliberate provision of preformed long-chain omega-3. None of these failures require malice. They arise from normal constraints: incomplete information, competing priorities, budget pressure, and the natural tendency to equate ethical purity with nutritional sufficiency.
A common example is the “whole-food vegan” household that refuses substitutes and fortification on principle. The household avoids fortified plant milks, avoids supplements, avoids products designed to close nutrient gaps, and relies on vegetables, grains, legumes, and fruit. This can look admirable and can provide many health benefits, but in child development it is often precisely the pattern that creates the highest logistics risk unless it is extraordinarily well executed. “No substitutes” is not automatically a sign of adequacy. It can be the opposite if it eliminates the very mechanisms by which vegan diets most reliably deliver B12 and other hard-to-secure nutrients. Another example is the household that uses supplements intermittently—purchasing them with good intentions but failing to sustain adherence across months of illness, travel, school routine changes, and caregiver variability. This again is not unintentional. It is a predictable failure mode of a regime that requires consistent execution but is run without systems.
Economic reality sits underneath this in a way that dietary debates often ignore. Engineering a developmentally adequate plant-only diet for a child is a higher-cognitive-load task than feeding a child a mixed diet that includes routine animal-source proteins. It requires knowledge, planning time, higher monitoring burden, and often higher per-unit cost for fortified foods and supplements. It also requires continuity: the diet must be executed by multiple caregivers in consistent ways. In households with limited time, limited money, or unstable routines, the “engineering overhead” competes with other necessities. This is not a moral judgement. It is a constraint model: when the marginal cost of correctness is high, error rates rise.
The cost is not only financial. There is an attention cost and a compliance cost. The household must know which nutrients are structurally risky in plant-only child diets, must know how to secure them, must purchase and store the relevant inputs, must track usage, and must maintain adherence despite a child’s variable appetite and preferences. The household must also resist drift into ultra-processed substitution patterns that can meet calories while undermining micronutrient density. These are not trivial tasks, and the difficulty is compounded by conflicting online advice, ideological pressure within communities, and the temptation to treat supplements as optional or “unnatural.” A mixed diet, by contrast, often solves multiple adequacy problems without requiring constant deliberation: not because it is inherently superior in every respect, but because nutrient co-packaging reduces the need for daily optimisation.
This is why the article treats “unplanned” as an economic and behavioural category as much as a nutritional one. Diets that require precision impose fixed costs: knowledge acquisition, time allocation, and compliance systems. Those costs are easier for high-resource households to pay. In affluent observational datasets, this can produce a misleading pattern where vegan or vegetarian groups appear to perform well because the households most likely to adopt the diet are also most capable of paying the engineering costs. That does not refute the thesis; it explains why adequacy must be measured directly rather than inferred from labels. The fragility is not uniform across households. It concentrates where resources, time, and monitoring capacity are scarce.
The article therefore uses “unplanned” as a disciplined shorthand: not a slur, not a claim about intent, but a description of a diet run without the explicit systems that make plant-only child nutrition reliably adequate. The empirical prediction follows directly. Where the engineering overhead is paid—fortified inputs are used, supplementation is consistent, and adequacy is monitored—height and IQ outcomes should look normal. Where the overhead is not paid—whether through cost constraints, time constraints, or principled refusal of fortification and supplementation—nutrient inadequacy becomes more common, growth constraints become more likely, and small but measurable cognitive penalties become more probable at the population level.: height-for-age z-score (HAZ) on the x-axis and IQ on the y-axis, with two diet groups overlaid as separate point clouds.
-
The purpose is diagnostic rather than decorative. If diet inadequacy is real, the “unplanned vegetarian” group is expected to show a left-shift in HAZ, a down-shift in IQ, and (in many datasets) a broadly positive within-group association between growth status and cognition. The fitted lines are included only to make the expected pattern legible: a higher-growth trajectory tends to align with higher cognitive performance when both are downstream of adequacy constraints.
-
In the article, this joint plot is used to separate diet identity from diet performance. A vegan/vegetarian subgroup with demonstrably adequate supplementation and protein quality should sit closer to the meat-access cluster than to the “unplanned” cluster.
The mechanism is simple. Growth is a cumulative process. Bone elongation, connective tissue development, and the expansion of lean tissues require continuous synthesis of structural proteins and continuous support from micronutrient-dependent pathways. When essential amino acids are marginal, the body cannot “average it out” indefinitely. When key micronutrients are intermittently absent—B12, iron, iodine, zinc, and other development-linked nutrients—the body does not simply compensate by eating more calories. It reallocates. It prioritises immediate survival and core function, and growth becomes the adjustable variable. This is why linear growth is often the first outwardly measurable sign of sustained nutritional constraint. It is not that a child becomes “ill” in an acute sense. It is that the child grows more slowly than expected over months and years. The constraint hardens into a trajectory.
Height-for-age is therefore the correct metric for the role it plays here. Height-for-age expresses growth relative to age norms, allowing the article to detect whether a child is tracking along an expected curve or drifting downward. Growth velocity adds an even sharper lens: it reveals whether the child is falling behind in real time rather than merely being short at one moment. A diet can look adequate in a cross-sectional snapshot if the child is short for familial reasons or if earlier adequate periods masked later inadequacy. Growth velocity and repeated height-for-age measures are the antidote to that confusion. They let the analysis distinguish stable trajectories from constrained trajectories.
This matters because the core argument is not “vegetarian” versus “meat” as identities. It is adequacy versus fragility. A diet can be vegan and developmentally adequate if it is engineered properly: adequate energy, adequate high-quality protein through deliberate combinations, consistent B12 provision, planned iron and zinc adequacy, deliberate iodine, and deliberate DHA/EPA provision. In such a case, the thesis does not predict that the child must be shorter. It predicts that when diet is adequate, growth should be normal. Conversely, a diet can be vegan in name and inadequate in reality: inconsistent supplementation, narrow food repertoire, high reliance on processed replacements, weak protein-quality planning, and recurring shortfalls during illness periods. In that case, the thesis predicts a higher probability of growth constraint. Height helps identify which world a child is in.
This is precisely why height is paired with IQ. IQ is a downstream outcome influenced by many variables and confounded by social and familial factors. If the analysis tries to infer diet adequacy from IQ alone, it risks being captured by confounding. Height provides an independent validation channel. If a diet pattern is truly delivering what development requires, it should support normal growth. If a diet pattern is failing on essential amino acids or critical micronutrients, linear growth is one of the most likely outcomes to show that failure. That does not mean height “causes” IQ. It means both height and cognition are downstream of adequacy constraints, and height is often the easier downstream indicator to measure reliably.
The pairing also disciplines the argument by forcing internal consistency. If the article claims that plant-only patterns are constraining neurodevelopment through amino-acid and micronutrient bottlenecks, but there is no sign of growth constraint and no sign of nutrient inadequacy markers, then the argument is weakened. If, however, a subgroup shows lower height-for-age trajectories, higher prevalence of deficiency markers or risk proxies, and lower cognitive outcomes, the argument gains a coherent pattern: the diet is constraining development across systems that share the same inputs. In other words, height prevents the analysis from drifting into pure story-telling. It forces a cross-check.
Height also helps separate two common confusions. The first confusion is to treat all vegetarian and vegan children as equivalent, when in reality diet planning quality varies dramatically. The second confusion is to treat vegetarian identity as a nutritional guarantee. It is not. A child can be raised as “vegan” and eat a diet dominated by refined carbohydrates and processed substitutes. That child can be calorie-sufficient and still developmentally constrained. Height-for-age and growth velocity are designed to detect exactly that kind of hidden inadequacy because they integrate dietary adequacy over time. They do not reward virtue signals. They report biology.
For the purposes of the article’s empirical plan, height therefore functions as an external validator of adequacy. It allows the analysis to discriminate between “diet label” and “diet performance.” A vegan child with excellent supplementation adherence and adequate protein quality should show growth trajectories that do not differ meaningfully from omnivorous peers, and if that subgroup also shows no cognitive penalty, the thesis is refined: the penalty is conditional on inadequacy, not on ideology. A vegan child whose diet is nutritionally inadequate should show a higher probability of constrained growth velocity and lower height-for-age, and if cognitive outcomes track in the same direction, the thesis is strengthened: the same adequacy failures that slow growth also impose pressure on neurodevelopment. That is why growth is the canary in this argument. It is not a moral measure. It is an external developmental readout.
7 — The confounding problem and how the article will handle it
Any article claiming that vegetable-protein-reliant child diets reduce IQ must confront the confounding structure head-on, or it becomes propaganda. In affluent populations, vegetarian status is not randomly assigned. It clusters with parental education, higher health literacy, lower smoking rates, different parenting routines, different school choices, and differences in home cognitive stimulation. Many of these variables are independently associated with child cognitive outcomes. That means a naïve comparison between “vegetarian children” and “omnivorous children” can easily produce misleading results in either direction. The very households most likely to adopt vegetarian patterns in high-income settings can also be the households that provide the strongest non-dietary cognitive advantages. Any argument that ignores this will be dismissed correctly.
This is why the article states its evidentiary burden plainly. If the claim is that vegetable proteins lower IQ in development, the effect must survive adjustment for the major confounders that plausibly drive both diet choice and child outcomes. At minimum, analyses must adjust for socioeconomic status, parental education, and household structure variables that are standard in developmental datasets. Where available, parental IQ proxies must be used, because parental cognitive ability is one of the strongest predictors of child IQ and it also correlates with dietary choices, health behaviours, and adherence to complex nutritional regimens. Birth weight and gestational age must be controlled because prenatal development influences later growth and cognition and can correlate with maternal diet and health behaviours. Home environment measures—such as reading frequency, childcare quality, educational resources, and indices of cognitive stimulation—must be included where possible because they represent direct non-dietary pathways into measured IQ. Total energy intake must be controlled because caloric sufficiency interacts with protein and micronutrient sufficiency, and because energy intake itself can differ by diet type and household routines.
The article will also treat confounding as a structural feature, not a nuisance. It is not enough to “control for SES” in a single model and declare victory. The correct approach is staged modelling that shows how estimates shift as covariates are added. The baseline model will compare outcomes by diet group. The next model will add socioeconomic and parental education variables. The next will add prenatal and early-life factors such as birth weight and gestational age. The next will add home environment measures. The next will add total energy intake and broader diet quality indices. This staged structure forces transparency: if the diet effect disappears when SES and parental variables are added, the result is not “a failure.” It is information. It suggests that the observed difference was primarily social selection rather than nutritional constraint. Conversely, if a diet effect persists after those adjustments, the result becomes more credible as a nutrition-linked effect.
Table 4. Confounders and controls: variables required to avoid a false conclusion-
Operational rule for the article: a claim that “vegetable proteins lower IQ” is treated as credible only if the estimated effect persists through staged adjustment across these domains and shows the predicted dependence on nutrient adequacy and supplementation adherence, rather than simply tracking the vegetarian label.
A second confounding issue is more subtle and more important for the thesis: “vegetarian” is not a biological exposure. It is a label that contains heterogeneous nutritional realities. In high-income settings, a well-planned vegetarian pattern can be associated with high diet quality, high fruit and vegetable intake, lower processed-food intake, and strong supplementation adherence. That constellation can improve health and can correlate with higher cognitive outcomes through multiple channels. This is why vegetarian children can look good in observational data. But a superficial positive association does not refute the thesis, because the thesis is not that vegetarian identity causes harm; the thesis is that developmental constraint emerges when adequacy fails, and adequacy failure is more likely when the diet requires precision and that precision is not maintained.
This is where “biological fragility” must be stated carefully. A well-planned vegetarian or vegan diet can be adequate. The fragility is not that it cannot work; the fragility is that it is compliance-dependent. The diet’s adequacy depends on consistent execution across time. In real households, compliance is not evenly distributed. It varies with parental education, conscientiousness, mental bandwidth, financial resources, and the practical realities of childcare. It varies across school environments, social events, and caregiver changes. It varies during illness periods when appetite collapses and dietary diversity narrows. A diet that is “adequate if engineered” therefore has a predictable distribution of outcomes: some households execute it very well; some do not; and the children in the lower tail of execution are the children the thesis predicts will show growth and cognitive penalties.
The article’s strategy is to stop treating labels as biology. It will separate diet groups into at least three categories: omnivorous with routine animal-source protein access, vegetarian (lacto-ovo) with regular dairy and/or eggs, and vegan (plant-only). But the crucial move comes next: within each category, the analysis will quantify nutrient adequacy and supplementation adherence rather than assuming it. Vegan without reliable B12 provision is not classified as “a vegan child diet” in a neutral sense; it is classified as “nutrient inadequate,” because B12 is not optional. Similar adequacy scoring will be applied to iron risk proxies, iodine sourcing, zinc adequacy proxies, and DHA/EPA provision. Protein-quality proxies will also be scored using diet composition patterns, including the diversity and pairing of protein sources and indicators of limiting amino-acid risk. The analysis will therefore compare not only diet identities, but adequacy strata: engineered adequate plant-only patterns versus unplanned plant-only patterns, and both versus animal-source access patterns.
This approach yields a claim that is both stronger and more precise than culture-war slogans. The test is not “vegetarian children versus omnivorous children.” The test is whether children exposed to plant-only patterns that fail adequacy criteria show measurable penalties in height trajectories and IQ outcomes, and whether those penalties persist after the confounders that make vegetarian households systematically different are properly handled. If the penalty vanishes when adequacy is secured, the thesis is refined: the causal exposure is not “plants,” it is inadequacy. If the penalty persists even when adequacy markers are satisfied, the thesis must be revised. Either way, the article earns credibility by treating confounding as a central design constraint rather than as an afterthought.
8 — Empirical tests and analysis plan (methods in prose)
A defensible demonstration of the thesis requires data that separate diet identity from diet adequacy and that treat height and cognition as outcomes measured under comparable conditions. The required dataset is therefore not simply a list of “vegetarian” children and “omnivorous” children. It is a dataset that includes repeated or reliable measures of diet pattern, supplementation behaviour, growth, and cognitive outcomes, with enough background variables to address confounding. The analysis plan below specifies what must be measured, how groups are defined, how outcomes are standardised, and what tests will be used to determine whether vegetable-protein-reliant patterns produce measurable penalties in height and IQ, particularly where diets are unplanned.
The sample must include children assessed within clear age bands, because both height and cognitive test performance are strongly age-dependent and because mixing ages produces artefactual differences. The article will use age bands that correspond to typical testing windows and schooling stages, such as early childhood (2–5 years), mid-childhood (6–10 years), and early adolescence (11–14 years), with a requirement that each band has adequate sample size in each diet group. If the dataset includes multiple cohorts or repeated measures across time, each cohort will be analysed separately first, then combined using meta-analytic or pooled modelling with cohort indicators to avoid silently averaging incompatible contexts. Sex will be recorded and used in modelling, not because the thesis is sex-specific, but because both height norms and cognitive distributions differ by sex and must be handled explicitly.
Diet classification rules must be operational, not rhetorical. At minimum, children are classified into three diet identity groups: omnivorous (routine animal-source protein access), vegetarian (lacto-ovo, with eggs and/or dairy consumed routinely), and vegan (plant-only, excluding animal-source foods). Routine access must be defined in behavioural terms using food frequency criteria: a diet is not omnivorous because a child ate meat once at a birthday party; it is omnivorous if animal-source proteins appear as a routine component of meals over a defined period. Similarly, vegan status requires a consistent plant-only pattern over the defined period. Because families can drift in and out of patterns, the classification window must be explicit (for example, the past month for contemporary diet, plus a longer history variable where available). Diet identity alone, however, is not treated as the causal exposure in this article. The causal exposure is adequacy. Therefore, within each identity category the analysis will construct an adequacy classification using a pre-specified checklist: consistent B12 provision (fortified foods and/or supplementation), protein-quality proxies (diversity and pairing sufficient to reduce limiting amino-acid risk), and coverage of key micronutrients relevant to development (iron, iodine, zinc, and DHA/EPA provision via diet or supplements). This produces a second-layer classification: engineered adequate versus unplanned/high-risk. The thesis predicts that penalties will cluster in the unplanned/high-risk subgroup, especially within vegan and plant-heavy patterns.
Exclusion criteria will be limited to factors that make height or IQ measurement non-comparable rather than factors that conveniently “clean” results. Children with conditions that directly and independently constrain growth (for example, severe chronic diseases known to alter growth trajectories) will be handled by pre-specified rules: either excluded if the aim is to estimate typical dietary effects, or analysed separately as a sensitivity analysis. Similarly, children with acute illness at the time of measurement that plausibly depresses test performance will be flagged, and analyses will be repeated with and without those observations. Prematurity and low birth weight will not be excluded by default; they will be adjusted for explicitly because they are common and developmentally relevant. Extreme implausible values, where present, will be handled transparently with pre-specified outlier rules and sensitivity checks rather than ad hoc deletion.
Figure 7. Analysis pipeline: raw diet data → classification → nutrient adequacy → outcomes → adjusted models-
This figure lays out the article’s analysis pipeline as a sequence of explicit steps:
-
Raw diet data (food frequency questionnaires and/or food diaries) plus recorded supplement use are first converted into diet identity categories (omnivore, vegetarian, vegan). Those labels are then not treated as the causal exposure by themselves. Instead, the pipeline constructs an adequacy score based on the hypothesised bottlenecks (B12, iron, iodine, zinc, DHA/EPA, plus protein-quality proxies), producing “engineered adequate” versus “unplanned/high-risk” strata.
-
Outcomes are then analysed in parallel—height-for-age z-score (HAZ), stunting prevalence, and IQ composites—followed by staged adjusted models that add confounders in steps and test whether any diet-identity association is attenuated when adequacy markers are introduced. The central design choice, shown at the bottom, is that adequacy is treated as the pathway variable, and the core prediction is a gradient: the largest penalties should cluster where supplementation adherence and adequacy markers fail.
Outcome measures are defined in two domains. Height outcomes are measured using height-for-age z-scores (HAZ) as the standardised primary endpoint, with growth velocity where repeated height measures exist. Stunting prevalence will be computed using the standard operational definition selected in advance. Cognitive outcomes will be measured as age-normed IQ composites or validated cognitive battery composites. The analysis requires that IQ is standardised within age bands and test versions. If multiple instruments are used across cohorts, each instrument will be standardised internally so that differences reflect group comparisons rather than instrument scaling differences. Where the dataset contains subtests, both overall composites and domain scores may be analysed, but the primary cognitive endpoint is a general composite to avoid cherry-picking subdomains.
Nutrient markers are included where available because they allow the analysis to test the proposed mechanism rather than relying on labels. At minimum, the analysis will attempt to include B12 status proxies (direct serum measures or functional proxies where present), iron indices (e.g., ferritin or equivalent markers where available), iodine adequacy proxies (use of iodised salt, dietary iodine estimates, or thyroid-related proxies if recorded), and omega-3 intake proxies (dietary intake estimates, supplement usage, or biomarker proxies if available). If biomarkers are not present, the analysis will use structured intake and supplementation variables as adequacy proxies, but will treat such analyses as weaker than biomarker-informed models.
The descriptive outputs are designed to show both centre and distribution. For each age band and cohort, the article will present histograms and density overlays for IQ by diet identity and by adequacy strata, along with group means, standard deviations, and effect sizes in IQ points. For height, it will present distributions of HAZ and stunting prevalence by group, along with mean HAZ differences. The joint-outcome plot (HAZ vs IQ) will be produced for each cohort and age band, with groups overlaid, because it visually tests the coherence of the adequacy constraint model: if diets are limiting development, both outcomes should move together in the predicted direction, with the most constrained subgroup clustering in the lower-left region. These descriptive plots are not treated as proof, but they are treated as essential diagnostics that prevent the analysis from being reduced to a single coefficient.
Inferential testing will use regression models with staged adjustment to address confounding and to expose where any estimated diet effect comes from. The baseline model estimates the association between diet group and outcomes (IQ and HAZ) controlling only for age band and sex. The next model adds socioeconomic status variables. The next adds parental education variables. The next adds prenatal and early-life variables (birth weight, gestational age). The next adds home environment measures where present. The next adds total energy intake and broad diet quality indices. Finally, models add adequacy variables and nutrient markers. This staged structure makes the argument falsifiable: if the “vegetable proteins lower IQ” coefficient collapses to zero once SES and parental education are included, the claim is social selection rather than nutrition. If it persists and is attenuated specifically when nutrient markers are included, that supports the pathway claim: diet identity is acting through nutrient adequacy. If the effect persists even when adequacy markers are satisfied, the thesis must be revised.
A mediation-style logic will be applied explicitly and cautiously. The hypothesis is that plant-only patterns depress height and IQ primarily through nutrient inadequacy and lower effective protein quality. Therefore, the analysis will test whether (i) diet identity predicts adequacy failures, (ii) adequacy failures predict outcomes, and (iii) adjusting for adequacy markers reduces or removes the diet identity association. This is not presented as definitive causal mediation unless the data support it with sufficient structure, but it is presented as a transparent pathway test rather than a narrative.
Table 5. Reporting template (populated per cohort and age band)
Assumptions used for the data: IQ is approximately normal within age bands; low-IQ threshold = 85, high-IQ threshold = 130; stunting = HAZ < -2; and “unplanned/high-risk” plant-only patterns have higher rates of adequacy failure, producing both lower mean outcomes and worse tail proportions. -
How this table is used in the article: one block is completed per cohort per age band, with rows for each diet identity group split by adequacy stratum where possible. The “Primary model covariates included” column documents the base specification, while “Staged adjustment result” reports the final diet effect after full controls. The “Sensitivity checks” columns prevent cherry-picking by forcing disclosure of robustness outcomes.
Robustness checks are built into the plan. Results will be compared across cohorts, across instruments, and across age bands. Sensitivity checks will test whether results depend on outliers, whether results change when using quantile-based comparisons rather than means, and whether results differ when excluding children with unreliable diet reporting. Critically, supplementation adherence will be treated as a sensitivity hinge: analyses will be repeated stratifying vegan/vegetarian groups by documented adherence to B12 and other key supplements. The thesis predicts a gradient. The strongest penalties should appear where plant-only patterns coincide with inadequate supplementation and inadequate nutrient markers; penalties should be reduced or absent in engineered adequate plant-only patterns. If the gradient does not appear—if engineered adequate plant-only patterns still show penalties of similar magnitude—then the thesis must be amended. If the gradient appears consistently, the thesis moves from assertion to measured claim.
9 — What would count as confirmation or failure
This article is not written to win an argument by volume. It is written to make a claim that can be confirmed or falsified with transparent criteria. The thesis is a developmental constraint thesis: diets that rely heavily on vegetable proteins, particularly plant-only patterns, impose a higher risk of nutrient inadequacy during childhood unless supplementation and planning are consistently strong, and those inadequacy states create measurable pressure on height trajectories and on IQ outcomes. Confirmation and failure must therefore be defined in terms of patterns that follow from the mechanism, not in terms of whether one group feels morally validated.
Confirmation requires a repeatable three-part pattern observed across cohorts, age bands, and instruments. The first part is an adequacy pattern. Plant-only or plant-heavy patterns must show a higher prevalence of nutrient inadequacy markers or inadequacy risk proxies than diets with routine animal-source nutrient co-packaging, unless supplementation is robust. The emphasis is on “unless,” because the thesis is not that vegan diets are intrinsically deficient; it is that deficiency risk rises when the diet’s adequacy is compliance-dependent and compliance is imperfect. In practical terms, confirmation requires that when vegan/plant-heavy households are stratified by demonstrated B12 provision, iodine sourcing, iron and zinc risk proxies, and DHA/EPA provision, the inadequacy subgroup is larger than in omnivorous groups, and the engineered adequate subgroup can be identified and separated cleanly.
The second part is a growth pattern. In the inadequacy subgroup, height outcomes must show a measurable constraint signal: lower mean height-for-age z-scores, higher prevalence of stunting or sub-threshold growth metrics, and/or reduced growth velocity where longitudinal data exist. This does not require catastrophic differences. The mechanism predicts drift and accumulation. A small reduction in growth velocity sustained across years is enough to produce a meaningful height-for-age shift. Confirmation requires that this growth constraint signal is not evenly distributed across all vegetarian/vegan children, but concentrates where inadequacy markers are present. If an engineered adequate vegan subgroup shows normal growth, while an unplanned/high-risk subgroup shows a downward shift in HAZ and/or growth velocity, that is mechanism-consistent evidence.
The third part is a cognitive pattern, and this is where the argument must be disciplined. Confirmation requires small but detectable IQ penalties in the inadequacy subgroup that persist after staged adjustment for confounding variables. “Persist” here does not mean “unchanged.” It means that after controlling for socioeconomic status, parental education, parental cognitive proxies where available, prenatal factors (birth weight, gestational age), home environment measures, energy intake, and overall diet quality, an independent association remains between inadequacy-linked dietary patterns and lower cognitive outcomes. The thesis also makes a stronger, pathway-specific prediction: when nutrient adequacy is demonstrably secured—through robust supplementation adherence and adequate nutrient markers—the IQ penalty should attenuate substantially or disappear. That attenuation is a key confirmation criterion because it is the hallmark of a nutrient-pathway mechanism rather than a purely social label effect.
These confirmation criteria can be summarised as a gradient requirement. Confirmation is not “vegans have lower IQ.” Confirmation is: the most constrained outcomes cluster where plant-only patterns coincide with demonstrated inadequacy; engineered adequate plant-only patterns do not show the same penalties; and the pattern replicates across cohorts and age bands.
Failure is equally clear and must be stated without evasions. The thesis fails if plant-only or plant-heavy patterns do not show systematic differences in adequacy markers, once measurement is reasonably competent. If nutrient inadequacy is not more prevalent in unplanned plant-only patterns than in omnivorous patterns, then the proposed logistics risk is not supported. The thesis also fails if there is no growth signal: if height-for-age and growth velocity show no measurable constraint in the inadequacy subgroup, the claim that development is being materially limited loses its external validator. The thesis further fails if there are no cognitive differences after adjustment, or if any apparent differences are fully eliminated by confounder control in a way consistent with social selection rather than nutrition.
A more direct failure mode is possible and must be admitted: if plant-only groups outperform across height and IQ outcomes independent of nutrient adequacy markers, the mechanism is wrong or incomplete. That pattern would imply that either the adequacy markers are not capturing the relevant biology, or that other correlated factors dominate outcomes in a way that the model cannot account for. Either way, it would not be honest to maintain the thesis in the face of such results.
In short, confirmation is a replicable adequacy–growth–cognition chain with attenuation under demonstrable adequacy. Failure is the absence of that chain, or an opposite pattern that remains after adequacy is measured.
10 — Conclusion: the claim stated without slogans
This article advances a developmental constraint thesis, not a lifestyle insult. The claim is that child diets that rely primarily on vegetable proteins—especially plant-only patterns—carry a higher probability of failing the nutrient logistics required for maximal growth and neurodevelopment unless they are deliberately engineered and consistently supplemented. That probability difference is the core of the argument. It is not that plant foods are harmful. It is that child development is time-sensitive, and the nutrients most tightly linked to development are not delivered automatically in plant-only patterns. Adequacy can be achieved, but it is conditional on execution.
The mechanism is straightforward. Protein quality is not merely protein grams; it is essential amino-acid adequacy and usable protein after digestion, and growth and brain development draw continuously from the same pool. Micronutrients are not decorative; they are co-factors and substrates that sit upstream of growth and neurodevelopmental pathways. Vitamin B12 is non-negotiable in vegan diets without fortification or supplementation. Iron, iodine, and zinc are common bottlenecks when diets are narrow or absorption is not considered. Preformed DHA/EPA is a further logistics hurdle when animal-source foods are excluded. These are not ideological points. They are constraints that must be solved repeatedly for years during development.
The article’s empirical structure is designed to keep the claim honest. It does not treat diet labels as biology. It separates vegetarian, vegan, and omnivorous patterns, then stratifies each by measured adequacy and supplementation adherence, and it uses height alongside IQ because growth is an external validator of internal adequacy. Where plant-only diets are engineered and supplementation is consistent, the thesis does not predict systematic penalties. Where plant-only diets are unplanned and adequacy fails, the thesis predicts a measurable pattern: growth constraint signals, higher prevalence of inadequacy markers, and small but detectable downward pressure on IQ distributions that persists after confounder adjustment and attenuates when adequacy is secured.
The final position is not anti-plant. It is pro-development. It treats height and IQ as measurable outcomes that track biological inputs, and it argues that children pay the price when adults confuse dietary identity with nutritional adequacy.
Addendum Prologue — Terms, execution realities, and the economics of “unplanned”
In this article, “unplanned” is a technical term about execution, not a claim about intent. It does not mean accidental, careless, or unaware. It means the diet is not run as an engineered system with explicit targets, verified inputs, and routine monitoring. A household can be intensely committed to veganism and still be “unplanned” in the relevant sense if it does not consistently secure the nutrients that vegan child development requires. Conversely, a household can be indifferent to dietary ideology and still be “planned” if it reliably hits the biological constraints through deliberate food choice, fortification, supplementation, and follow-through.
The article therefore uses a set of distinct terms to avoid confusion. “Diet identity” refers to what foods are excluded or included: omnivorous, vegetarian (lacto-ovo), and vegan (plant-only). “Diet adequacy” refers to whether the pattern meets developmentally required inputs over time. “Engineered adequate” refers to diets that meet adequacy criteria consistently—adequate energy density, adequate effective protein (amino-acid completeness plus digestibility), and deliberate management of the micronutrients that do not reliably appear in plant-only patterns. “Unplanned/high-risk” refers to diets that are not managed with that engineering discipline, whether because the household lacks information, time, money, continuity of caregivers, or because the household believes that “whole foods” automatically solve the logistics. The term is chosen because the risk is structural: when adequacy requires repeated precision, a diet without systems is statistically more likely to drift into shortfall.
This is why the article repeatedly separates “unplanned” from “unintentional.” Many vegan households intend the best possible outcome. The problem is that intent does not generate B12, iodine, or DHA/EPA. Implementation does. An unplanned vegan pattern is often characterised by one or more of the following: inconsistent or absent B12 supplementation; reliance on food choices that are nutritionally virtuous but not micronutrient-dense; avoidance of fortified products because they are perceived as “processed”; lack of an iron or zinc strategy; absence of iodised salt as a stable iodine source; and no deliberate provision of preformed long-chain omega-3. None of these failures require malice. They arise from normal constraints: incomplete information, competing priorities, budget pressure, and the natural tendency to equate ethical purity with nutritional sufficiency.
A common example is the “whole-food vegan” household that refuses substitutes and fortification on principle. The household avoids fortified plant milks, avoids supplements, avoids products designed to close nutrient gaps, and relies on vegetables, grains, legumes, and fruit. This can look admirable and can provide many health benefits, but in child development it is often precisely the pattern that creates the highest logistics risk unless it is extraordinarily well executed. “No substitutes” is not automatically a sign of adequacy. It can be the opposite if it eliminates the very mechanisms by which vegan diets most reliably deliver B12 and other hard-to-secure nutrients. Another example is the household that uses supplements intermittently—purchasing them with good intentions but failing to sustain adherence across months of illness, travel, school routine changes, and caregiver variability. This again is not unintentional. It is a predictable failure mode of a regime that requires consistent execution but is run without systems.
Economic reality sits underneath this in a way that dietary debates often ignore. Engineering a developmentally adequate plant-only diet for a child is a higher-cognitive-load task than feeding a child a mixed diet that includes routine animal-source proteins. It requires knowledge, planning time, higher monitoring burden, and often higher per-unit cost for fortified foods and supplements. It also requires continuity: the diet must be executed by multiple caregivers in consistent ways. In households with limited time, limited money, or unstable routines, the “engineering overhead” competes with other necessities. This is not a moral judgement. It is a constraint model: when the marginal cost of correctness is high, error rates rise.
The cost is not only financial. There is an attention cost and a compliance cost. The household must know which nutrients are structurally risky in plant-only child diets, must know how to secure them, must purchase and store the relevant inputs, must track usage, and must maintain adherence despite a child’s variable appetite and preferences. The household must also resist drift into ultra-processed substitution patterns that can meet calories while undermining micronutrient density. These are not trivial tasks, and the difficulty is compounded by conflicting online advice, ideological pressure within communities, and the temptation to treat supplements as optional or “unnatural.” A mixed diet, by contrast, often solves multiple adequacy problems without requiring constant deliberation: not because it is inherently superior in every respect, but because nutrient co-packaging reduces the need for daily optimisation.
This is why the article treats “unplanned” as an economic and behavioural category as much as a nutritional one. Diets that require precision impose fixed costs: knowledge acquisition, time allocation, and compliance systems. Those costs are easier for high-resource households to pay. In affluent observational datasets, this can produce a misleading pattern where vegan or vegetarian groups appear to perform well because the households most likely to adopt the diet are also most capable of paying the engineering costs. That does not refute the thesis; it explains why adequacy must be measured directly rather than inferred from labels. The fragility is not uniform across households. It concentrates where resources, time, and monitoring capacity are scarce.
The article therefore uses “unplanned” as a disciplined shorthand: not a slur, not a claim about intent, but a description of a diet run without the explicit systems that make plant-only child nutrition reliably adequate. The empirical prediction follows directly. Where the engineering overhead is paid—fortified inputs are used, supplementation is consistent, and adequacy is monitored—height and IQ outcomes should look normal. Where the overhead is not paid—whether through cost constraints, time constraints, or principled refusal of fortification and supplementation—nutrient inadequacy becomes more common, growth constraints become more likely, and small but measurable cognitive penalties become more probable at the population level.
References
Lemale, J., Mas, E., Jung, C., Bellaiche, M., Tounian, P., & Association Française de Pédiatrie Ambulatoire. (2023). Diet and growth of vegetarian and vegan children. BMJ Nutrition, Prevention & Health, 6(Suppl 2), s3–s10. (BMJ Nutrition)
Neumann, C. G., Murphy, S. P., Gewa, C., Grillenberger, M., & Bwibo, N. O. (2007). Meat supplementation improves growth, cognitive, and behavioral outcomes in Kenyan children: A randomized controlled trial. The Journal of Nutrition. (ScienceDirect)
Northstone, K., Joinson, C., Emmett, P., Ness, A., & Paus, T. (2012). Are dietary patterns in childhood associated with IQ at 8 years of age? A population-based cohort study. Journal of Epidemiology & Community Health, 66(7), 624–628. (Jech)
Schürmann, S., Kersting, M., & Alexy, U. (2021). Nutrient status and growth in vegan children. Nutrition Research Reviews. (ScienceDirect)
Smithers, L. G., Golley, R. K., Mittinty, M. N., Brazionis, L., Northstone, K., Emmett, P., & Lynch, J. W. (2013). Do dietary trajectories between infancy and toddlerhood influence IQ in childhood and adolescence? Results from a prospective birth cohort study. PLOS ONE, 8(3), e58904. (PLOS)
Petersen, K. S., Burrows, T. L., & Collins, C. E. (2024). Plant-based diets in children: Secular trends, health outcomes, and a roadmap for future research. Nutrients, 16(5), 723. (MDPI)
Influence of a vegan diet on child health and development: A scoping review. (2025). Nutrition Reviews. (academic.oup.com)