23 août 2025

Childhood tastes and veggie choices: what the evidence shows

In the British Gemini twin cohort, genetic factors account for 60 % of food fussiness at 16 months and 83 % by age 3. Most of what separates a child who tastes everything from one who refuses nearly everything does not come from how the parents fed them.

That is an uncomfortable number, because it cuts against the familiar story about childhood tastes: a family table writing a lifelong food programme. The reality is more layered and, for someone trying to eat more plants at forty, more useful. The levers are real. They are just smaller and more specific than the usual advice suggests.

What prenatal exposure actually changes

The claim that a fetus tastes its mother’s diet is not a metaphor. It rests on a randomised trial published in Pediatrics in 2001, and the details of what it measured are worth reading closely.

Pregnant women planning to breastfeed were assigned to three groups. One drank 300 ml of carrot juice four days a week for three weeks in the final trimester, then water during the first two months of lactation. A second did the reverse. A third drank water throughout. About four weeks after solids were introduced, and before the infants had ever eaten anything containing carrot, they were fed cereal prepared with water at one session and carrot-flavoured cereal at another.

Infants exposed to carrot, either in amniotic fluid or in breast milk, showed fewer negative facial expressions with the carrot cereal than with the plain version. Unexposed infants showed no such difference. Mothers in the prenatal exposure group also rated their babies as enjoying the carrot version more.

The detail that matters comes next. The same tendencies appeared for the amount of cereal eaten and the length of the feeds, but those findings were not statistically significant. So the study demonstrates real sensory learning, measured through facial expressions and maternal perception. It does not demonstrate that the child will eat more carrots.

A systematic review with meta-analysis in PLOS One in 2023 pooled twelve studies on this chemosensory continuity between prenatal life and the first postnatal year. Infants do orient toward odours they encountered before birth. The mechanism is established. It is also modest.

The Gemini study followed 4804 British twins born in 2007. Parents rated food fussiness at 16 months, 3, 5, 7 and 13 years using a validated scale.

The trajectory is the same for all children: fussiness rises from 16 months to age 7, then declines slightly between 7 and 13. That peak around seven is worth knowing, because it means the point at which a child seems most difficult is usually the top of a curve rather than the start of a decline.

The heritability figures are more striking. At 16 months, genetic factors explain 60 % of the differences between children, with a confidence interval of 53 to 67 %. By age 3 that share reaches 83 %, interval 81 to 86 %. Every parameter of the trajectory sits under strong genetic influence, above 70 %.

Environment is not absent, but it splits in a particular way. Shared environment, meaning everything twins in one household experience in common, only has a significant effect during toddlerhood. Environment unique to each child explains 15 to 26 % of the variance over time, with some influence persisting from age 5 onward.

That structure has a direct consequence. If you and your sibling grew up at the same table and one of you loathes spinach while the other asks for more, it is not an anomaly. It is the expected result. The shared table carries less weight than we give it.

A second finding from the same study is easy to miss. The genetic influences overlap substantially across every measurement point, meaning largely the same underlying factors operate at 16 months and at 13 years. Food fussiness behaves less like a sequence of separate phases and more like a continuous trait that varies in intensity. For parents that carries one practical message: a child who rejects a lot at three rarely becomes the opposite at eight, though they usually become somewhat easier.

Repeated exposure: a real effect, and a small one

The most widely repeated advice about childhood tastes is that a food must be offered many times before it is accepted. The principle holds. The magnitude does not match the enthusiasm.

A 2020 Cochrane review pooled 80 randomised trials, 218 trial arms and 12,965 participants aged five and under. Fifty of those trials examined child-feeding practices such as repeated exposure.

The result: compared with no intervention, these practices produce a small positive effect on vegetable consumption, equivalent to 5.30 grams of additional vegetables. The authors rate that as low-quality evidence. Multicomponent interventions, which combine things like parent nutrition education with changes to childcare policy, produce an equally small effect equivalent to 0.34 cups of fruit and vegetables a day, on moderate-quality evidence. Parent nutrition education on its own gives an uncertain effect on very low-quality evidence.

Five grams of vegetables is half a mouthful. Mean follow-up after the intervention was 8.3 weeks, and very few trials measured anything past six months. Of the 80 trials, only 23 were free from high risk of bias across all domains.

A cluster-randomised trial in Norway shows the same ceiling. Forty-six kindergartens were split between a control group and two intervention groups served a hot lunch with three rotating vegetables, the second group also receiving weekly sensory lessons. The sensory group ate more of the intervention vegetables. The effect on total vegetable intake was inconclusive, and no effect at all was observed on food neophobia.

None of this makes repeated exposure pointless. It remains the best-documented lever available. But a lever that shifts five grams is not a transformation method; it is a practice to sustain over years without expecting a turning point.

What the Norwegian trial separates is also worth noting. The effect appeared for the three vegetables actually served, not for vegetable intake overall. Repeated exposure works on the specific food rather than on general openness. If you want a child to eat fennel, you have to offer fennel; practising with carrots does not transfer on its own.

Pressure to eat follows the child rather than shaping them

One recent reversal deserves attention from anyone raised on the instruction to clean their plate.

A 2025 study in the International Journal of Behavioral Nutrition and Physical Activity followed 588 Chinese children, mean age 3.7 years, across three waves over two years using cross-lagged panel models. Those models separate two questions that usually get conflated: do families who apply pressure have fussier children, and does pressure at one point in time precede a worsening in that same child?

On pressure to eat, both models converge on a child-driven effect. Food fussiness predicts later increases in parental pressure, not the reverse. The authors conclude that pressure to eat and restriction function more as reactions to the child’s behaviour than as parent-initiated strategies.

This does not make pressure to eat advisable. It changes how an adult reads their own history. If you remember a table where you were pushed to eat, the likeliest explanation is not that the pressure created your difficulty. It was probably also a response to it.

What taste genetics says, and what it does not

The TAS2R38 gene is routinely presented as the reason some people hate broccoli. A phenome-wide analysis published in the European Journal of Nutrition in 2025 lets us size that claim.

The study examined three TAS2R38 variants against 139 food-liking traits and 29 food-intake traits, using UK Biobank data covering up to 445,779 people of European ancestry.

The bitter-sensitive alleles were associated with reduced preference or consumption for horseradish, salty food, grapefruit and alcohol, and for alcohol only whisky, red wine and spirits. They were also associated with increased preference for cucumber and melon, and higher tea consumption.

That list is short and specific. Across 139 liking traits, the effect concentrates on a handful of intensely bitter or pungent items. Inherited taste sensitivity nudges particular preferences. It does not sort people into vegetable lovers and vegetable refusers.

What this means for an adult

Nearly all the research above studies children. Carrying it over to an adult calls for caution, and this is exactly where most advice goes wrong.

What holds: familiarity increases acceptance at any age, through a learning mechanism that does not switch off after childhood. A food tasted once and rejected has not been properly tested.

What does not hold: the notion that an adult can rewrite their preferences within weeks given the right technique. The effects measured in children, whose preference systems are the most plastic, are small and observed over a few weeks.

The practical implication runs almost opposite to the challenge mindset. Rather than aiming to convert your tastes, it is more realistic to build meals around what you already like and add slowly. Our page on the ideal pantry for a beginner vegetarian starts from that premise, and the one on five common mistakes when going vegetarian describes what happens when the transition is forced.

One point deserves stating plainly, because it usually gets lost in talk of pleasure and memory: liking a food and digesting it comfortably are different things. A sudden jump in legume intake produces discomfort that has nothing to do with taste, and which will nonetheless be read as dislike. Our page on soaking and sprouting legumes covers the technical side.

Childhood dishes, adapted without illusions

Turning a family dish vegetarian works, for a specific reason: it reduces novelty. The format, the sauce, the texture and the setting stay recognisable, and only the central ingredient changes. It is the familiarity principle that makes repeated exposure work, applied in a single step.

A few adaptations built on that logic rather than on straight substitution:

A shepherd’s pie with green lentils keeps the mash, the browned top and the gravy. Green lentils hold their shape through cooking and keep a granular texture close to minced meat.
A meat-free chilli works because the dish’s aroma profile comes from long-cooked cumin, chilli and tomato rather than from meat.
A quiche with plant cream and silken tofu keeps the pastry and the sliceable set of the filling, which are the two recognisable elements.
A vegetable soup blended with a potato recovers the smoothness of childhood soups without cream.

The most effective adaptation is not the most faithful one but the one that preserves whatever you personally associate with the dish. For some people that is the sauce, for others the browned crust. Our guide to replacing meat in your favourite recipes works through substitutions by the technical role played in the recipe, and the complete guide to plant proteins gives the matching portions.

Feeding a vegetarian child: what the professional bodies say

If the question concerns your own children, it stops being about preferences and becomes about monitoring.

The nutrition committee of ESPGHAN, the European society for paediatric gastroenterology, hepatology and nutrition, published a position paper in 2025 based on a systematic search. Ten articles were retained, covering roughly 1500 children on a vegan diet, plus three systematic reviews with meta-analysis.

Their conclusion is measured. Current evidence is inconclusive as to whether a strictly vegan diet supports normal childhood growth, although no significant differences in height or body mass index z-scores were observed compared with omnivorous peers. The committee recommends regular monitoring of intake, growth and nutritional status, particular attention to protein, omega-3, calcium and iron, and vitamin B12 supplementation where the diet is strictly vegan.

It matters what that document covers. It addresses the vegan diet, which is more restrictive than a vegetarian diet retaining eggs and dairy. Our pages on vitamin B12 and on plant iron absorption deal with the two nutrients most involved. On the practical side, our page on getting children to enjoy vegetables collects the mealtime habits that follow from this evidence.

In Switzerland, the FSVO published in 2025 the scientific basis for nutritional recommendations for children and adolescents aged 4 to 17, drawn from peer-reviewed systematic reviews and meta-analyses published between January 2010 and March 2025. Until that work, Swiss recommendations for this age group remained largely undifferentiated, resting mainly on the Swiss Society for Nutrition’s food disc.

Common mistakes

Concluding after one try. A food tasted once, in one preparation, on one occasion, has not been evaluated. This is the point on which the evidence converges most clearly.

Expecting repeated exposure to flip a preference. The documented effect is around five grams of vegetables, on low-quality evidence over a few weeks of follow-up.

Blaming a child’s fussiness entirely on how they were raised. By age 3, 83 % of the differences between children trace to genetic factors, and shared environment only matters significantly during toddlerhood.

Reading the peak of difficulty around age 7 as deterioration. The measured trajectory rises to that point and then falls back through to 13.

Applying pressure at the table. Longitudinal models suggest the practice follows a child’s fussiness rather than correcting it.

Mistaking digestive discomfort for dislike. A rapid increase in fibre and legumes produces symptoms that get attributed to taste.

Putting a child on a vegan diet without monitoring. The ESPGHAN paper calls for growth surveillance and vitamin B12 supplementation.

By situation

For an adult changing how they eat at forty, plasticity is lower than in childhood but autonomy is far higher. Choosing the preparation, the context and the quantity yourself offsets part of the acquired rigidity.

For a vegetarian parent, most of the work sits in what gets offered and in calm repetition, not in negotiation at the table. Offering without requiring remains the practice best supported by the evidence, modest as its effect is. Organisation counts for more than persuasion, as our page on vegetarianism in a large family sets out.

For someone whose childhood meals were a battleground, it helps to know the pressure was probably as much a consequence as a cause. That does not change the memory, but it changes the conclusion drawn from it.

For anyone who knows they are highly bitter-sensitive, the TAS2R38 findings point somewhere concrete: the measured effect concentrates on a few intensely bitter foods. Preparations that cut bitterness, such as roasting to caramelisation, act on the parameter actually involved.

For a household where one person turns vegetarian, familiarity cuts both ways. An adapted shared dish meets less resistance than an entirely new dish served alongside. Festive meals concentrate that tension, and our page on family gatherings as a vegetarian handles that case, as does the one on answering common objections.

What this evidence does not say

The carrot juice trial covers a small sample and a single flavour. It establishes a mechanism, not a dietary strategy for pregnancy.

The Gemini heritability figures describe how differences are distributed within one British population at one point in time. A heritability of 83 % does not mean 83 % of a child’s behaviour is genetically fixed, nor that environment has no effect on that child.

The Cochrane review itself stresses the low quality of much of the evidence and the near-complete absence of follow-up beyond six months.

The pressure-to-eat study covers 588 Chinese children with a mean age of 3.7 years. Family feeding practices vary considerably between cultural settings.

The TAS2R38 analysis covers people of European ancestry and measures statistical associations without establishing a causal mechanism for each food listed.

The ESPGHAN paper rests on ten articles and states explicitly that well-designed prospective research is still needed.

Frequently asked questions

Can you really learn to like a vegetable you have hated since childhood? Acceptance rises with familiarity at any age. The measured effects are real but small, and varying the preparation counts as much as repeating it.

How many times should a food be offered to a child? Trials use protocols of several successive exposures, but no solid data fixes a universal number. The Cochrane review measures the overall effect of these practices, not a threshold.

My child is fussier than before, should I worry? The average trajectory rises until age 7 before falling. Worsening between 2 and 6 matches the expected pattern. If the diet narrows enough to affect growth, that is a question for the paediatrician.

Will eating varied food during pregnancy make my baby less fussy? Prenatal exposure alters responses measured at birth and during weaning. No data supports promising a less selective child.

Is a taste for meat cultural or biological? Both contribute, and the available studies cannot quantify their respective shares for a food this culturally loaded.

Can a child be vegetarian safely? A vegetarian diet including eggs and dairy is easier to balance than a vegan one. For the latter, ESPGHAN calls for growth monitoring and vitamin B12 supplementation.

Why do my siblings and I have such different tastes? That is the expected outcome. Shared environment explains only a small share of the differences, and mainly during toddlerhood.

Should vegetables be hidden in dishes? Hiding them raises intake without building acceptance of the food itself, because it removes the sensory exposure that creates familiarity.