The first definition of ultraprocessed is "industrial food formulations made up mostly or entirely of ingredients... that are not found in a similar form and combination in nature". That includes a lot of merely 'processed' foods. A highly aged cheese will have basically no naturally-occurring ingredients left, and any bread with leavening and refined flour fails "form and combination found in nature". Similarly, the fat and salt flavors of strong cheese, or the sweetened flavors of a pastry with sugar, go beyond "sharpening taste" to creating entirely new nonnatural flavors.
We can all apply the Potter Stewart test to say that Twinkies are ultraprocessed and sharp cheddar isn't, so I don't meant to be pedantic here. But I think this loose definition points to genuinely important unresolved questions.
Where on the scale from sausage to Twinkie does the problem start?
Is it really true that an Entenmann's coffee cake causes vastly different eating habits than equal access to a home-made version? Can we isolate the difference?
Above all, which differences actually drive this? 'Ultraprocessed' is not a food additive but a loose class of recipes, and it'd be nice to reduce that to a distinction we could put on a label.
The problem probably doesn't isolate to any one additive or alteration, but we should be able to find something more concrete than a vague naturalistic appeal. The study in question has some promising work in that direction, like controlling for energy density. Outright added calories are a long-standing suspect (since we're likely to use the same amount of e.g. spaghetti sauce despite store-bought versions having far more sugar), but Hall's work suggests that the problem persists even without that. I'd very much like to see more of these studies to replicate the effect and extract a more substantive definition of 'ultraprocessed'.
