Your body resists weight loss hard and weight gain barely#
The resistance is real, and if you have held a weight below your usual one you have felt both halves of it: hunger that outlives the diet, and a body burning less than its new size predicts. What is not established is the thing the phrase "set point" quietly smuggles in — a specific defended number, encoded somewhere in the brain, that your physiology is steering back toward. Nobody has located that number in a person, and three competing models explain the same observations without agreeing that it exists.
The disagreement is live and unusually specific: researchers have named the experiment that would separate the models, run it twice, and gotten opposite answers. Everything below is that argument. The one fact all three models must account for is an asymmetry — the defense against losing weight is strong and reproducible, while the defense against gaining is weak enough that most of a population could drift upward for four decades without visibly hitting it. This spoke of why weight-loss plateaus happen is about what that asymmetry implies. The measured size of the downward defense belongs to adaptive thermogenesis, and the argument at the level of the hormone is settled in leptin resistance.
Three models of the same body#
A 2011 review assembled by two dozen obesity researchers laid out the options and, notably, did not pick one1. Two decades of the field's disagreement fit into three sentences.
The set point model says adiposity is actively regulated around a reference value: drift above it and intake falls while expenditure rises; drift below it and the reverse happens until you are back. The settling point model says there is no reference value at all — weight simply settles wherever intake and expenditure passively balance in a given environment, the way water level settles in a leaky tank. The dual intervention point model splits the difference in a way neither of the others can: there are two independent thresholds, a lower and an upper, with an unregulated zone of indifference between them. Inside the zone, weight drifts more or less at random. Outside it, physiology intervenes.
| Model | The claim | Prediction after weight loss | Where it struggles |
|---|---|---|---|
| Set point | Active feedback defends a target level of fatness | Reactive hyperphagia — you eat more than your old baseline until fat is restored | Explaining why a defended number would shift upward across a whole population in 40 years |
| Settling point | No target; weight settles where intake and expenditure balance | Intake returns to baseline and weight passively drifts back | Explaining why weight loss is resisted so vigorously at all |
| Dual intervention point | Two independent thresholds with an unregulated zone between them | Intake returns to baseline; resistance appears only below the lower threshold | Nobody has located either threshold in a living person |
One complication belongs here, because it argues that the machinery is plural rather than singular. Implanting weighted capsules into rats and mice reversibly reduced their body weight through lower food intake, relieved diet-induced obesity, and improved glucose tolerance — and the effect vanished in mice depleted of osteocytes, implicating a load sensor in the weight-bearing bones rather than anything in fat tissue. The authors called it a gravitostat and read it as a second negative-feedback system for fat mass, independent of leptin4. That is rodent work, not human physiology, and it drew immediate published argument. But it makes the point that "the" set point may be shorthand for several controllers with different sensors.
The experiment that separates the models, run twice, with opposite answers#
Here is the clean discriminator, and it is worth understanding because it decides which story about your own body is true.
When a dieter stops dieting, what happens to intake? If a set point is being defended, you should see reactive hyperphagia: eating above your original pre-diet baseline, not merely back at it, and staying there until fat mass is restored. If instead the system is only correcting an energy imbalance — the settling-point and dual-intervention-point reading — intake should climb back to baseline and stop, with weight drifting up passively from there. The predictions differ, and they differ in a measurable quantity.
Semaglutide withdrawal supplied the test twice. In one trial, patients with obesity lost weight over 17 months and then came off the drug; the modelled energy-balance dynamics showed a substantial degree of reactive hyperphagia — the set-point signature. In another, patients took semaglutide for five months and were then randomized to placebo or continuation for a further year; the placebo group raised intake back to baseline levels without overshooting, which is the dual-intervention-point signature. Reviewing both, Speakman and Hall concluded that the contradiction supports neither model cleanly and points instead at psychological factors that neither model accommodates well2.
That is the current state of the science, and it is worth stating plainly rather than resolving: the two trials differed in duration, in design, and in how intake was inferred rather than measured, and no one has yet run the deliberate head-to-head. A 2025 review of the field opened by noting that models of body weight regulation exist but experiments testing their validity are lacking3. Anyone telling you set point theory is established, or that it has been debunked, is ahead of the data in one direction or the other.
Where a defended number runs into 1980#
The set-point model's hardest problem is not a laboratory result. It is a date. Body weight in most wealthy countries rose substantially from around 1980 onward, which a system actively defending a target is not supposed to permit. The available patch is to say the set points themselves drifted upward — but as Speakman and Hall put it, we know nothing of the molecular basis of a set point, so the reason that might have happened is unclear2. A mechanism invoked only to absorb the anomaly is a weak mechanism.
The dual-intervention-point model takes the epidemic more comfortably: people gained without physiological resistance because they were inside the unregulated zone, and they differ from one another in where their upper threshold sits. Speakman's evolutionary reading of why an upper limit would be loosely specified is that predation once selected for one — an animal too heavy to escape gets eaten — and that when humans invented tools, weapons and fire, roughly two million years ago, the selective pressure holding the upper point in place lapsed and random mutation was free to scatter it.
That is a hypothesis, not a finding, and it should be read as one. What is measured is the scatter itself: identical twins overfed by 84,000 calories under supervision gained between 4.3 and 13.3 kg, with co-twins resembling each other far more than they resembled other pairs — the genetic spread in how people handle a surplus is large and reproducible. Whatever the upper limit is, it is not the same limit for everyone.
What you can act on while the models argue#
The practical content survives the theoretical dispute, because all three models agree on the asymmetry even where they disagree about its cause. Going down is defended. Going up is barely defended. That has three consequences worth acting on.
First, treat a stall near your lightest sustainable weight as information rather than failure — telling that apart from an ordinary breakable plateau has its own procedure, in plateau or set point. Second, stop looking for a number to reset. There is no measured set point to raise or lower, so the advice to "fix your set point" with a food, a supplement or a protocol is selling a target no one has found. Third, put the effort where the asymmetry says the pressure is: maintenance, not the diet. If your physiology pushes hardest at the bottom, the phase that needs a plan is the one after the weight is off — which is exactly what the maintenance evidence shows in practice.
What you can move is the environment side of the ledger, which every model grants matters, and where the settling-point view is at its strongest. The weight you settle at is partly a function of what is in the fridge, how far you walk, and how much you sleep — inputs with no defended value attached to them at all.
FAQ#
Does your body have a fixed weight it wants to return to?#
Not one anybody has found. Three competing models — set point, settling point, and dual intervention point — explain the same observations, and a 2011 review by two dozen researchers declined to choose between them1. What is well established is the asymmetry: resistance to weight loss is strong and reproducible, while resistance to weight gain is weak. "Your body defends a range, downward much harder than upward" is a defensible statement. "Your body defends 78 kg" is not.
Can you reset your set point?#
The question assumes a quantity that has never been measured in a person, so nothing can be shown to reset it. The falsifiable version — does the resistance weaken if you hold a lower weight for long enough? — is exactly what the field says it lacks experiments for3. What is documented is that sustained maintenance is achievable and has identifiable habits behind it, whatever the underlying control system turns out to be.
If the body defends its weight, why did obesity rise so quickly after 1980?#
This is the set-point model's weakest point and the dual-intervention-point model's strongest. A system actively defending a target should have resisted a population-wide rise; the rescue — that the targets themselves drifted upward — has no known molecular basis2. A model with an unregulated zone between two thresholds predicts exactly what happened: people drifted upward without resistance until they reached their own upper limit, and those limits differ widely between individuals.
Sources#
- Speakman JR, Levitsky DA, Allison DB, Bray MS, de Castro JM, Clegg DJ, et al. Set points, settling points and some alternative models: theoretical options to understand how genes and environments combine to regulate body adiposity. Dis Model Mech. 2011;4(6):733-45.
- Speakman JR, Hall KD. Models of body weight and fatness regulation. Philos Trans R Soc Lond B Biol Sci. 2023;378(1888):20220231.
- Fernández-Verdejo R, Ravussin E, Galgani JE. Body weight regulation models in humans: insights for testing their validity. Nat Rev Endocrinol. 2025;21(11):703-717.
- Jansson JO, Palsdottir V, Hägg DA, Schéle E, Dickson SL, Anesten F, et al. Body weight homeostat that regulates fat mass independently of leptin in rats and mice. Proc Natl Acad Sci U S A. 2018;115(2):427-432.
- Bouchard C, Tremblay A, Després JP, Nadeau A, Lupien PJ, Thériault G, et al. The response to long-term overfeeding in identical twins. N Engl J Med. 1990;322(21):1477-82.
