一日三餐是谁定的？
晚上十一点，写字楼的灯还亮着一半。外卖骑手在楼下排成一列，便利店的关东煮热气腾腾，夜市的摊子刚刚支起来。对今天的许多年轻人来说，这才是他们一天里的第一顿正经饭—早晨那一顿，睡过去了；中午那一顿，凑合了。
老一辈看见这一幕，多半要摇头：一日三餐，老祖宗传下来的规矩，怎么能乱？年轻人的回答也很干脆：饿了再吃，不更顺应身体吗？
两边吵的其实是两个问题：吃几顿，和几点吃。过去二十年，研究者把这两笔账都认真算了一遍。算完发现，两边都只说对了一小半。
先说三餐的身世。一日三餐并不是刻在基因里的。在古罗马，体面人一天只吃一顿正餐；农耕时代的欧洲人，多半一天两顿，跟着日头走。把吃饭钉成早、中、晚三格的，是工业革命的工厂铃声：上工前一顿，午休一顿，收工后一顿，吃饭第一次被排进了别人的时刻表。至于「早餐是一天中最重要的一餐」，据食物史学者考证，这句话最早是二十世纪初麦片厂商的广告语，是卖出来的，不是研究出来的。
三餐是作息制度的产物。这个出身决定了：它值得尊重，但不是生理定律。
那么，实验室怎么看待「次数」之争？方法很直接：把总热量锁死，一组吃三顿，一组吃六顿，看谁瘦得多、代谢更好。2015 年《营养学评论》汇总 15 项随机对照试验的荟萃分析发现，少食多餐表面上的那点优势，在剔除一项异常研究后消失了—总热量相同，餐次多少对减重和代谢率没有可靠影响。2024 年《美国医学会杂志》子刊汇总 29 项随机试验、2485 人，结论是吃得少顿的一组平均多减约 1.85 公斤，但作者自己写明：效应小，临床意义不确定，多数试验偏倚风险高。还有一项常被引用的试验，2014 年发表于《糖尿病学》：54 名 2 型糖尿病患者，同样的热量，吃两顿（早餐加午餐）的一组，减重、肝脏脂肪和胰岛素敏感性都优于吃六顿的一组。
这些研究合起来指向同一句话：决定胖瘦的首先是总量。餐次排在很后面，远没有两派吵的那么重要。
真正的分水岭，是钟点。这门学问有个名字，叫时间营养学。西班牙穆尔西亚大学的加劳莱特（Marta Garaulet）团队跟踪了三千三百多名参加减肥计划的人：晚吃的人比早吃的人少减重约 17%。另一项针对一万九千名女性的研究显示，晚饭吃到临睡前两小时以内、或者饭后还要吃零食的人，超重和肥胖的比例明显更高。约翰斯·霍普金斯大学的对照实验更直观：同一顿饭，晚上十点吃比傍晚六点吃，餐后血糖峰值高约两成，夜间脂肪燃烧少约一成。2025 年一项双胞胎研究进一步发现，一个人若把一半以上的热量吃在自身生物钟的后半段，胰岛素敏感性要低约三成—与吃多少无关，只与什么时段吃有关。
这正是年轻人生活方式的要害。晚睡晚起本身不是罪，少吃的那顿早餐也不是关键；要命的是整个进食窗口被一路推向深夜，再叠上晚宴和宵夜，一天里的大部分热量压进了身体最不擅长处理它们的时段。研究者把这种身体钟与社会钟的错位叫作「社会时差」。夜里十一点那顿饭之所以让人不安，不是因为它违规，而是因为它在替整个错位的节奏买单。
再说「饿了再吃」。这一派在学术界有个正式名字，叫直觉进食：不看钟点、不守餐次，靠身体的饥饱信号决定吃与停。支持它的证据不弱。明尼苏达大学 Project EAT 项目调查了 2287 名平均 25 岁的年轻人：自称信任身体信号的人，体重指数更低，暴食、长期节食等进食紊乱行为也更少；同一项目的八年纵向跟踪得出了相同方向的结果。2021 年一项汇总 97 项研究的荟萃分析同样显示，直觉进食得分高的人，进食障碍症状更少，自我形象更好。
但这派证据有一个共同的软肋：绝大多数是观察性研究。它们证明了「会听身体话的人更健康」，却证不了是前者导致了后者—也可能恰恰相反，是生活规律、没被节食折腾过的人，才保得住一套灵敏的饥饱信号。更要紧的是，饥饿感并不总是可靠：熬夜、压力、长期节食，都会让它失灵；该响的时候不响，不该响的时候乱响。
这一点，在厌食症的病房里看得最清楚。进食障碍的国际标准疗法、牛津大学研发的强化认知行为疗法（CBT-E），治疗的第一步恰恰是把「饿了再吃」反过来：按钟点规律进食—每天三顿正餐加两到三次加餐，间隔不超过约四小时，不管饿不饿，先照表吃。英国国家卫生与临床优化研究所的指南也是同样的思路。道理不复杂：对这类患者，饥饿信号本身就是病的一部分，等信号来才吃饭，等于不吃。要先用外部的节律把身体喂稳，内部的信号才有机会慢慢修回来。
于是同一个问题出现了两种答案：在代谢基本健康的成年人那里，吃几顿、守不守点，弹性很大，总量和时段对了就行；在信号已经失灵的人那里，规律不是束缚，是治疗。这也是为什么这场争论吵不出赢家—两边说的根本不是同一种人。
回到夜里十一点的餐桌。科学没有给现代人发一本新规矩，它只是把顺序排了出来：总量第一，时点第二，规律性第三，次数排在最后。三餐不是天条，它是工厂时代的遗产；饿了再吃也不是万能钥匙，它对信号完好的人成立，对信号失灵的人危险。
人这一辈子吃饭的方式，总是被生活先安排一遍：先是工厂的铃，后是写字楼的灯，现在轮到手机屏幕和深夜的外卖软件。我们真正要问的，或许不是一天该吃几顿，而是—在被排满的日程里，还能不能给身体留一段它自己的时间。
（记者注：本文为健康议题专题报道，汇总公开研究结论，不构成个体诊疗建议。进食障碍、糖尿病等代谢疾病患者，请以临床医生意见为准。）
Who Invented Three Meals a Day?
At eleven at night, half the office lights are still on. Delivery riders queue downstairs, the convenience-store oden pot is steaming, and the night-market stalls are just setting up. For many young people today, this is the first proper meal of their day. Breakfast was slept through. Lunch was improvised.
The older generation shakes its head at the scene. Three meals a day is the rule handed down for generations — how can it be broken? The young answer is just as crisp: eat when you are hungry. Is that not closer to what the body wants?
The argument is really about two questions: how many meals, and at what times. Over the past two decades, researchers have carefully audited both. The result: each side is only a little bit right.
Start with the ancestry of the three-meal day. It is not written into our genes. In ancient Rome, respectable citizens ate one main meal a day. In agrarian Europe, most people ate twice, following the sun. What pinned eating into three fixed slots — morning, noon, night — was the factory bell of the Industrial Revolution: a meal before the shift, a meal at the midday break, a meal after work. Eating was scheduled by someone else's timetable for the first time in history. As for "breakfast is the most important meal of the day," food historians trace the phrase to cereal advertising in the early twentieth century. It was sold, not discovered.
Three meals a day is a product of the work schedule. That origin matters: it deserves respect, but it is not a law of physiology.
So what does the laboratory say about meal frequency? The method is simple: lock total calories in place, feed one group three meals and another six, and see who loses more weight and whose metabolism improves. A 2015 meta-analysis in Nutrition Reviews, pooling 15 randomized controlled trials, found that the apparent edge of frequent small meals vanished once a single outlying study was removed. With calories held equal, how often you eat has no reliable effect on weight loss or metabolic rate. A 2024 analysis in JAMA Network Open, pooling 29 trials and 2,485 people, found the lower-frequency group lost about 1.85 kilograms more — but the authors themselves called the effect small, of uncertain clinical importance, with most trials at high risk of bias. And in a widely cited 2014 trial in Diabetologia, 54 patients with type 2 diabetes eating the same calories did better on two meals, breakfast and lunch, than on six: more weight loss, less liver fat, better insulin sensitivity.
Taken together, the trials point to one sentence: total intake decides first. Meal frequency comes far behind — nowhere near as important as the two camps claim.
The real dividing line is the clock. The field has a name: chrononutrition. Marta Garaulet's team at the University of Murcia followed more than 3,300 people through a weight-loss program: late eaters lost about 17 percent less weight than early eaters. In a study of more than 19,000 women, those whose dinner fell within two hours of bedtime, or who snacked after dinner, were markedly more likely to have overweight or obesity. A controlled experiment at Johns Hopkins made it vivid: the identical meal eaten at 10 p.m. produced blood-sugar peaks about 20 percent higher and overnight fat burning about 10 percent lower than the same meal eaten at 6 p.m. A 2025 twin study went further: people who consumed half their calories in the later half of their own circadian day had insulin sensitivity roughly 30 percent lower — regardless of how much they ate. Timing, not quantity.
That is precisely where the young lifestyle takes its hit. Sleeping late is not the sin, and the skipped breakfast is not the point. The damage is that the whole eating window slides toward midnight, and with late dinners and night snacks, most of the day's calories land in the hours when the body handles them worst. Researchers call the mismatch between the body clock and the social clock "social jet lag." The unease about that 11 p.m. meal is not that it breaks a rule. It is that it pays the bill for an entire shifted rhythm.
Now the other side: eating when hungry. In the literature it has a formal name, intuitive eating — no clock, no meal plan, letting hunger and fullness signals decide when to start and stop. The supporting evidence is real. Project EAT at the University of Minnesota surveyed 2,287 young adults averaging 25 years of age: those who trusted their body's signals had lower body mass index and fewer disordered behaviors such as binge eating and chronic dieting. An eight-year longitudinal follow-up of the same project found the same direction. A 2021 meta-analysis of 97 studies likewise found that higher intuitive-eating scores came with fewer eating-disorder symptoms and better body image.
But this body of evidence has a shared weakness: most of it is observational. It shows that people who can read their body's signals are healthier. It cannot show which caused which. It may be the reverse — that people with orderly lives, never battered by dieting, are the ones whose hunger signals survived intact. And hunger is not always a reliable instrument. Short sleep, stress and a history of restriction can all break it: silent when it should ring, ringing when it should not.
Nowhere is that clearer than in an eating-disorder ward. The international standard treatment, enhanced cognitive behavior therapy (CBT-E), developed at Oxford, begins by reversing "eat when hungry" entirely: eat by the clock. Three planned meals plus two or three planned snacks, never more than about four hours apart, whether you feel hungry or not. Britain's National Institute for Health and Care Excellence recommends the same. The logic is plain: for these patients, the hunger signal is part of the illness. Waiting for it before eating means not eating. An external rhythm has to feed the body steadily before the internal signal can slowly repair itself.
So one question yields two answers. For metabolically healthy adults, frequency and punctuality are flexible, as long as total intake and timing are roughly right. For people whose signals have already failed, regularity is not a constraint — it is the treatment. That is why the argument never produces a winner: the two sides are not talking about the same people.
Back to the table at eleven at night. Science has not issued modern life a new rulebook. It has simply ranked what matters: total intake first, timing second, regularity third, meal count last. Three meals a day is not a commandment; it is a legacy of the factory age. Eating when hungry is not a master key either; it works for intact signals and endangers broken ones.
The way a person eats is always arranged by life first: once by the factory bell, then by the office light, now by the phone screen and the midnight delivery app. The better question may not be how many meals a day we owe ourselves, but whether, inside a schedule packed by everything else, we can still leave the body a stretch of time that belongs to it.
(Editor's note: This feature summarizes published research on a health topic and does not constitute individual medical advice. People with eating disorders, diabetes or other metabolic conditions should follow their clinician's advice.)
发布：GSM·美国（发布链接）
图片来源：图：一桌家常饭菜，桌上一碗洗净的红提。图片来源：James 提供。 / Photo: A home-cooked family dinner with a bowl of red grapes. Image provided by James.
GSM Global · 本篇英文版 English edition of this story
