Toddlers who have a diet high in processed foods may have a slightly lower IQ in later life, according to a British study described as the biggest research of its kind.
The conclusion, published on Monday, comes from a long-term investigation into 14,000 people born in western England in 1991 and 1992 whose health and wellbeing were monitored at the ages of three, four, seven and eight and a half.
Parents of the children were asked to fill out questionnaires that, among other things, detailed the kind of food and drink their children consumed.
Three dietary patterns emerged: one was high in processed fats and sugar; then there was a "traditional" diet high in meat and vegetables; and finally a "health-conscious" diet with lots of salad, fruit and vegetables, pasta and rice.
When the children were eight and a half, their IQ was measured using a standard tool called the Wechsler Intelligence Scale.
Of the 4000 children for which there were complete data, there was a significant difference in IQ among those who had had the "processed" as opposed to the "health-conscious" diets in early childhood.
The 20 per cent of children who ate the most processed food had an average IQ of 101 points, compared with 106 for the 20 per cent of children who ate the most "health-conscious" food.
"It's a very small difference, it's not a vast difference," said one of the authors, Pauline Emmett of the School of Social and Community Medicine at the University of Bristol.
"But it does make them less able to cope with education, less able to cope with some of the things in life."
The association between IQ and nutrition is a strongly debated issue because it can be skewed by many factors, including economic and social background.
A middle-class family, for instance, may arguably be more keen (or more financially able) to put a healthier meal on the table, or be pushier about stimulating their child, compared to a poorer household.
Emmett said the team took special care to filter out such confounders.
"We have controlled for maternal education, for maternal social class, age, whether they live in council housing, life events, anything going wrong, the home environment, with books and use of television and things like that," she said.
The size of the study, too, was unprecedented.
"It's a huge sample, it's much much bigger than anything anyone else has done," she said in an interview with AFP.
Emmett said further work was needed to see whether this apparent impact on IQ persisted as the children got older.
Asked why junk food had such an effect, she suggested a diet that was preponderantly processed could lack vital vitamins and elements for cerebral development at a key stage in early childhood.
"A junk food diet is not conducive to good brain development," she said.
The paper appears in the peer-reviewed Journal of Epidemiology and Community Health, published by the British Medical Association.