"It's willpower"

At the bottom of the bag, the verdict is willpower. You haven't got any. Other people can have three crisps and put the bag away, and you can't, and that's a fact about you.

Two findings, and both are about what willpower is up against.

David Marchiori and colleagues, in 2014, asked whether the portion size effect works the way anchoring does in other judgements, where a number you're shown first pulls your estimate toward it. They had people imagine being served a small or a large portion and say how much they'd eat. Those who imagined a large portion said they'd eat more than people given no portion to imagine, and those who imagined a small one said less. Then they told half the participants that the portion they'd imagined was random and meant nothing, and it made no difference. The anchor held. This was a study of imagined eating, and I wouldn't take the numbers to the sofa. But the mechanism is the one to notice. The size of what's in front of you sets how much you'll eat, and knowing that doesn't undo it.

Amar Cheema and Dilip Soman, in 2008, tested partitions. They gave people chocolates either in one quantity or divided into smaller units, and found that people ate less when the quantity was partitioned. The partition worked by drawing attention to each decision to continue, and the researchers found something else that matters here. The effect appeared when people were trying to control their eating and disappeared when they weren't. Partitions are a tool for people who want to eat less than the bag. They don't do anything to people who want the bag.

Put those together. A sharing bag is an anchor, and it pulls the amount you'll eat toward a hundred and fifty grams, and knowing it's an anchor doesn't help. Willpower is you, in the moment, trying to stop a hand that's being pulled by a bag, with a screen on, and no information about how far you've got. That's not a fair fight and it's not a test of character. It's a test of whether one person can beat a unit, an anchor and a distraction at once, and the answer for nearly everyone is no.

The partition is what wins, and it wins for a reason. A bowl turns one decision, open the bag, into two, fill the bowl and then decide whether to get up. Cheema and Soman's participants controlled their eating better with partitions because the partitions gave them somewhere to decide. Willpower was never the tool. The bowl is, and it works precisely for the person who wants to have a few, which is you.