The All-or-Nothing Diet Trap: Why Restriction Always Leads to Overeating

It usually begins on a Monday. The cupboards have been cleared on Sunday evening, the meal plan printed, and the new rules are absolute: no bread, no sugar, no snacks after 7pm. By Wednesday lunch the system is intact. By Friday, after a colleague's birthday cake, the structure has cracked. By Sunday — the day before the next Monday — the kitchen has been raided and the diet officially abandoned. A week later the loop restarts.

This sequence is not a personal failing. It is the most reliably documented pattern in behavioural eating research, and it has a name. Janet Polivy and Peter Herman, working at the University of Toronto in the 1980s, called it the "what-the-hell effect." Alan Marlatt, studying relapse in addiction recovery at the University of Washington, called it the "abstinence violation effect." Different labs, different patient populations, the same phenomenon: a single broken rule triggers a disproportionate collapse of the rule system itself.

The 1985 study that named what every dieter knows

The classic Polivy and Herman experiment was elegant in its cruelty. They recruited restrained eaters — women who routinely tried to limit food intake — and unrestrained eaters as controls. Half of each group was given a milkshake "preload" before being told they were participating in a taste test of several ice creams. They could eat as much as they wanted.

The unrestrained eaters who had drunk the milkshake ate less ice cream than the unrestrained eaters who had not. Their bodies had registered the calories and adjusted. The restrained eaters showed the opposite pattern. After the milkshake — which they perceived as a violation of their dietary rules — they ate significantly more ice cream than the restrained eaters who had skipped it. The diet had already been broken, so the brakes came off entirely.

Polivy and Herman published the finding in the Journal of Abnormal Psychology and built a theoretical framework around it: cognitive restraint creates a dichotomy between "on the diet" and "off the diet," and once the "on" boundary is crossed, the cognitive control that was holding eating below biological hunger collapses. Eating then accelerates well beyond what the original violation accounted for.

Why the milkshake mattered more than its calories

The key detail is that the preload's actual caloric content was irrelevant. In follow-up studies, researchers told restrained eaters they had consumed a high-calorie milkshake when it was actually low-calorie, and vice versa. The effect tracked perception, not biology. What disinhibited eating was the belief that the rule had been broken, not the metabolic state the food had produced. The trigger was cognitive, not nutritional.

This matters because it locates the all-or-nothing pattern not in the food, the hunger, or the willpower, but in the structure of the rules themselves. Rigid rules generate rigid violations.

Marlatt's abstinence violation effect

Around the same time, Alan Marlatt was studying why people who had stopped drinking, smoking, or using drugs frequently lapsed back into heavy use after a single slip. His work, published in 1985 in the foundational text Relapse Prevention, identified what he called the abstinence violation effect: when a person committed to abstinence violates the rule, they experience guilt, attribution of the lapse to a personal flaw, and a sense that the recovery effort has been definitively spoiled. The cognitive response to the slip — not the slip itself — predicts whether a small lapse becomes a full relapse.

Marlatt's framework transferred directly to eating behaviour. The single cookie does not biologically dismantle a week of careful eating. The mental response to the cookie — the framing of the day as ruined, the implicit decision to "start fresh tomorrow" — is what produces the binge that follows.

This pattern is also why the biology of dieting compounds the psychology. After several weeks of caloric restriction, ghrelin is elevated, leptin is suppressed, and the homeostatic pressure toward eating is already strong. A rigid rule system collapsing under that pressure is not a moral event. It is the predictable interaction of two systems both pointing in the same direction.

Flexible restraint versus rigid restraint

In 1994, Joachim Westenhoefer at the University of Hamburg made a distinction that has held up in subsequent research for three decades. He proposed that cognitive restraint comes in two forms. Rigid restraint involves all-or-nothing rules, dichotomous thinking, and identity-level commitments ("I am a person who does not eat carbs"). Flexible restraint involves general guidelines, contextual judgement, and the ability to compensate gradually rather than punitively after a deviation.

His data — and the studies that have replicated it since — consistently find that rigid restraint correlates with higher body weight, more disinhibited eating, and worse maintenance outcomes. Flexible restraint correlates with the opposite. The people who keep weight off long-term are not the ones following the strictest rules. They are the ones who have absorbed the inevitable deviations into a broader pattern rather than treating each one as a system failure.

This is counterintuitive to anyone who has been told that the discipline is the answer. The discipline, in its rigid form, is part of what creates the loop.

Why "starting fresh on Monday" makes the pattern worse

The Monday reset has a specific behavioural function: it converts each lapse into a discrete failure event with a defined endpoint. The day, the week, the diet — all of these become things that can be "ruined" and then reset. The reset itself becomes a small reinforcement; it provides the cognitive relief of a clean slate, which makes the next abandonment slightly easier to perform because the next reset is also waiting.

What this structure precludes is a graduated response: eating slightly more than planned at lunch and adjusting modestly at dinner, without designating either meal as part of "on" or "off" the diet. The Monday-reset model has no language for that kind of midcourse correction. Everything is either intact or broken.

What happens when the biology is treated, not overridden

The all-or-nothing pattern is most pronounced when the underlying hunger system is fighting the dietary rules at every meal. When that pressure is reduced — by sleep, by structured high-protein eating, or by pharmacological treatment — the cognitive structure can loosen without immediate collapse, because the body is no longer pushing against it constantly.

This is part of what patients on GLP-1 receptor agonists often describe in clinical interviews. The mental energy previously spent maintaining rules is no longer needed in the same way, because the rules themselves matter less. Hunger has diminished. The reward pull from previously off-limits foods has attenuated. A piece of bread is no longer the symbolic event it was when it had to be defended against; it is just a piece of bread, and there is no special urgency to eat the rest of the loaf.

The mechanism does not eliminate the psychology — flexible restraint is still a skill, and rigid thinking patterns can persist even when hunger is reduced — but it removes the biological amplifier that turns a small dietary deviation into a four-hour eating episode. For patients setting realistic expectations for treatment, this is often the change that matters most in daily life, even more than the scale movement.

Practical implications

The research does not prescribe a specific way of eating. It does suggest that any approach built on rigid rules — whether keto, intermittent fasting, calorie counting, or eliminating specific food groups — will be vulnerable to the same collapse pattern unless the rules are explicitly designed to absorb deviations. The cognitive behavioural therapy literature for eating concerns has converged on a small set of principles that consistently improve outcomes: avoid all-or-nothing framing, plan for inevitable deviations rather than treating them as failures, separate the lapse from its meaning, and give up the Monday reset.

None of this is comfortable. The all-or-nothing approach has the appeal of moral clarity. Flexible restraint feels, initially, like permission to fail. The data suggest the opposite: it is the conditions under which sustained eating change actually becomes possible.

Key takeaways

• Polivy and Herman's 1985 study showed restrained eaters consume more after a perceived dietary violation — the cognitive trigger, not the calories, drove the disinhibition.
• Marlatt's abstinence violation effect predicts that a single lapse becomes a full relapse based on how the person interprets the lapse, not on the lapse's actual size.
• Westenhoefer's distinction between rigid and flexible restraint has held up for thirty years: rigid restraint correlates with worse weight outcomes; flexible restraint with better ones.
• The Monday reset converts deviations into discrete failure events, which reinforces the loop rather than breaking it.
• Reducing the underlying biological pressure — including via GLP-1 medications — allows the cognitive structure to loosen without immediate collapse, because hunger and reward pull are no longer constantly fighting the rules.
• Planning for inevitable deviations, separating the lapse from its meaning, and abandoning the all-or-nothing framing are the consistent recommendations of the cognitive behavioural literature.