Two questions, and they take about ten seconds.
One: would a plain meal do? If someone put a plate of eggs and toast in front of you right now, would you eat it? Two: how long has it been since you ate, and what was in that meal?
The first question separates hunger from a craving. The second tells you whether the craving has a fuel gap sitting underneath it. Neither answer is a verdict on your character, and neither one means the correct move is to eat nothing. If you want the practical end of this rather than the diagnostic end, the pairing rule is the single most useful thing on this site.
Question one: would a plain meal do?
This one is old folk knowledge, usually called the apple test, and I have never found a study behind it. I am telling you that up front because it gets presented as science and it is not. It survives because it works often enough to be worth ten seconds.
The logic is simple. Hunger is a general question, so almost any food answers it. Chicken and rice, or a bowl of soup. A craving is a specific request, and a specific request rejects substitutes. If the honest answer to eggs and toast is "no, but I would eat a cookie," you are not hungry in the way the word normally means.
Where it goes wrong: if you are properly, several-hours hungry, everything sounds good including the cookie, so both answers come back yes and the test tells you nothing. That is not a failure of the test. That is the test telling you to skip to question two.
Question two: when did you last eat, and what was in it?
Timing is the part people forget, and it is usually the part carrying the information.
If it has been more than four hours, or the last thing you ate was mostly carbohydrate with very little protein in it, your body has a real case. Blood sugar climbs after a carbohydrate-heavy meal, insulin comes to clear it, and you can land lower than where you started a couple of hours later. That dip is a physical signal, not a mood, and the food it asks for is fast sugar because fast sugar resolves it quickest.
If it has been under three hours and that meal had protein in it, the fuel argument is weak. Something else is driving this, and the most likely candidate is a cue: the time of day, the end of work, the specific chair, a screen, the walk past the kitchen. Cues create wanting without any shortage behind them, which is why a craving can arrive twenty minutes after a real meal and feel exactly as urgent.
The four answers
Plain meal yes, long gap. This is hunger. Eat food. Not a dessert-shaped negotiation, food. Anything else you do here is an attempt to solve the wrong problem, and it usually ends with the dessert plus the meal an hour later.
Plain meal yes, short gap. The last meal was too small, or too light on protein to hold. Eat something with protein in it now, and then treat the meal itself as the thing to change rather than treating tonight as a failure. A lunch that reliably wrecks your afternoon is a lunch problem.
Plain meal no, long gap. A craving sitting on top of a genuine fuel gap, and the fuel gap is why it feels so loud. Feed the gap first with something real, then decide about dessert from a calmer place. Often the dessert still happens, and it is smaller, and it is fine.
Plain meal no, short gap. A cue. You are not hungry and you want something sweet, and both of those are allowed to be true. This is the answer people find hardest, because there is no physiological excuse to hide behind and the temptation is to call it a discipline problem. It is a learned association. You can eat something, you can not eat something, and neither choice says anything about you.
What to do in either case
The unhelpful version of this advice ends at "so now you know it is a craving, therefore do not eat it." That is not what the test is for. It is for picking a response, not for issuing a ruling.
If it came back hunger, eat a meal. If dessert is happening anyway, put it after the meal rather than instead of it. Sugar on an empty stomach produces the steepest curve and the fastest return of the same craving, so the order genuinely matters even when the food does not change.
If it came back craving with a fuel gap, the pairing is doing double duty. Something with protein in it closes the gap and answers the want at the same time. Our cottage cheese ice cream is 180 calories a bowl with 22 grams of protein, which is roughly what three eggs carry, and it reads as dessert rather than as a nutrition exercise.
The pattern generalises well past our own recipes. Yogurt with something sweet stirred through it works. A bar you made on Sunday works. Anything that puts the sweet thing next to something slow will do the same job, whoever wrote the recipe. There are a dozen more of ours, with the real numbers printed, in the high-protein roundup.
If it came back cue, you have three options and all of them are legitimate. Have the thing. Have a smaller version of the thing. Or notice the cue, wait, and see whether it passes, understanding that this is the option most likely to end with you back in the kitchen at ten.
The one I would steer you away from is the fourth option nobody lists, which is deciding you will not have it and then having a great deal of it two hours later. Restriction reliably increases wanting. The more completely something is off limits, the more attention it takes up, and the bigger the eventual break tends to be.
The size answer, when the answer is "have it"
The reason "have it" is safe advice here is that having it does not have to mean making a pan.
Single-serve is the underrated fix for a cue-driven craving. You want a cookie, so make one cookie, and there is no tray sitting on the counter tomorrow morning quietly asking you the same question again. Our single-serve collection exists for exactly this: a mug cake or a deep-dish cookie in a ramekin, portioned by the recipe rather than by willpower at the counter.
The other version is the freezer. Something already made, already portioned, waiting. A craving that meets a plan is a much smaller event than a craving that meets an empty cupboard and a car key.
What the test does not do
It does not tell you whether you should eat. It sorts the signal so you can pick a response that actually matches it, which is a smaller and more useful job.
It also does not diagnose anything. If cravings are constant, or they are getting worse, or they come with symptoms that worry you, that is a conversation for a doctor rather than a blog. I write recipes. None of this is medical advice, and anyone setting targets for you should have seen your labs.
The short version, for the fridge
Would a plain meal do? When did you last eat, and did it have protein in it?
Yes and a long gap means eat. No and a short gap means it is a cue, and a cue is not a character flaw. Everything in between is some mix of the two, and the useful move in almost every case is the same one: put protein next to it, whatever "it" turns out to be. If added sugar is the line you are holding rather than protein, start from the no added sugar collection instead.
