Older toddlers - aged about 2 - are picking up on potentially harmful anti-fat attitudes from their mothers, research suggests.
That is a finding from a University of Otago study of 70 Dunedin infants and toddlers, and their parents.
The study, by researchers from New Zealand, Australia and the United States, followed other research showing obesity prejudice and discrimination were increasing.
The team showed the tots pairs of photos of people - one in which the person was obese, and the other in a normal weight range. Researchers also used questionnaires to gauge the mother's attitude to obesity.
Professor Ted Ruffman, of Otago's psychology department, said: "What we found is that younger infants, around 11 months of age, preferred to look at obese figures, whereas the older toddler group, around 32 months old, preferred to look at average-sized figures." And that preference was "strongly related to maternal anti-fat prejudice".
There was a high correlation - the more the mother had expressed anti-fat attitudes in the questionnaire, the more the older toddlers would "look away from the obese figure towards the normal weight one".
The professor said anti-fat prejudice was "associated with social isolation, depression, psychiatric symptoms, low self-esteem and poor body image".
He emphasised the research findings were not meant to be a "mother-blaming exercise", but it indicated "how early children begin to absorb and display the attitudes of those around them". Mothers tended to be the primary caregivers and were "just reflecting wider societal attitudes".
Previous research had indicated anti-fat prejudice could be seen in pre-school children aged slightly more than 3 and was well-established in 5- to 10-year-olds.
But research by Professor Ruffman and his team suggested these attitudes had even earlier beginnings, apparent from about 2 years 8 months. "The surprising thing is that the link is so early," he said.
Researchers had considered other potential factors, such as parental body mass index (BMI) - a measurement derived from weight and height - as well as education, and the amount of children's television viewing.
But these were found to be unrelated to the sort of figure the child preferred to look at.
These were clear findings, just published in the Journal of Experimental Child Psychology, and Professor Ruffman believed they would interest parents.
"It's always good to find these new things and to get them published."