There is a familiar article about this. It lists the things grandparents do on a visit that a child quietly files under duty rather than affection: the glance at the clock, the round of dutiful questions about school, the kiss that functions as a checkpoint. It is satisfying to read, and if you are a grandparent it is quietly alarming, which is part of why it circulates.
It also has no research behind it. Nobody has established a set of grandparent behaviors that grandchildren later encode as obligation. Lists of that kind are written from intuition, and intuition about other people’s childhoods is not a strong instrument.
What has been established is more interesting than a list, and it comes at the question from two directions at once.
The Artful Age
A weekly letter on aging well, family across generations, and the creative life after the kids leave home.
Children start checking earlier than anyone assumes
Michelle Eskritt and Kang Lee showed young children a video. In it, a girl tastes refreshments her friend has made and says, while the friend is standing there, that she likes them. Once the friend has gone, she says she does not.
The question was which version the watching child would treat as true.
Three-year-olds could not tell. Four- and five-year-olds, the researchers report, were significantly more likely to rely on the private expression than the public one. In a third experiment, the older children did better when the mismatch showed up in the girl’s manner rather than in her words.
The authors call the four-year-olds’ performance remarkable, on the grounds that earlier work had placed this understanding considerably later in childhood.
This needs holding at its actual size. It is a video about a girl and some snacks, not a study of grandparents, and the authors note that their scenario was an optimized one, always pairing public liking with private dislike, so how far it travels is unclear. What it does establish is narrow and relevant: quite small children already know that the polite version of a feeling and the real one can come apart, and they have a working rule for deciding which to believe.
What actually predicts anything
The second direction is longer range. Laura Pittman and colleagues asked 474 late adolescents about each of their living grandparents: how good the relationship was now, and how much contact there had been back in elementary school.
What they found was that relationship quality was linked to depressive symptoms, self-worth, perceived competence in close friendships and perceived competence in romantic relationships. Grandparent contact was not.
That is one study, in one country, and its full text sits behind a paywall, so it is worth being careful with. Except that the same pattern turned up again on the other side of the world. Li and Sun surveyed 371 young adults in Hong Kong, measuring contact frequency and emotional closeness both in childhood and now. When both went into the model together, they report, relationship quality had significant unique effects on psychological outcomes and contact frequency did not.
Two unrelated teams, two years apart, two very different samples, landing on the same distinction. Replication of that kind is not proof, but it is the thing that separates a finding from a fluke.
What this does not mean
It does not mean the visiting is pointless, and reading it that way would get the practical lesson exactly backward.
Pittman’s team also ran moderation analyses, and found that more contact during childhood strengthened some of the positive associations between relationship quality and later self-worth and confidence in close friendships. Contact multiplied what was already there. What it did not do was manufacture it.
Both studies are correlational, and neither can establish which way the causal arrow points. Li and Sun say so plainly, and add that self-report invites social desirability and errors of recollection, and that their participants came from a single university. Pittman’s team asked adolescents to recall how much contact they had in primary school, which is a long way to ask anyone’s memory to reach.
So what is the answer to the original question
There is no list. That is the honest answer, and it is less useful than the list would have been if the list were true.
What a grandchild registers as obligation does not appear to be a behavior that could be swapped out for a better one. It is the distance between the performance and whatever is running underneath it, and on the evidence above, children start measuring that distance somewhere around four, using the channel adults are worst at controlling.
This is uncomfortable precisely because a list would have been fixable. You could stop doing item three. You cannot stop doing the thing underneath item three by deciding to.
The consolation is in the second finding, and it is a real one. If the quality is what carries, then the arithmetic of grandparenting is not about frequency, and nobody needs to feel they are failing because the drive is three hours or the visits are quarterly. A grandparent who genuinely wants to be there, four times a year, is on the side of the evidence. So is the one who goes every Sunday and means it, and whose visits multiply something that was already true.
The visits are not the relationship. They are where the relationship becomes visible, to a smaller and more attentive person than most of us remember being.