“Exercise doesn’t help you lose weight” is one of the most shared lines in nutrition, and it usually comes with Herman Pontzer’s name attached. Pontzer, an evolutionary anthropologist at Duke University, did produce the research behind the idea that daily energy expenditure is constrained. But the slogan stretches that research well past what it shows. To see where, it helps to read the studies in order.
Study by study
2012: the Hadza
The starting point was a paper on the Hadza, hunter-gatherers in northern Tanzania who walk long distances every day to forage and hunt. Pontzer and colleagues measured their total daily energy expenditure with doubly labeled water, the field’s reference method: participants drink water tagged with stable isotopes, and the rate at which those isotopes leave the body reveals how much carbon dioxide, and therefore how much energy, they produced over a period of days.
The result surprised the team. Once differences in body size were taken into account, the Hadza burned roughly the same number of calories per day as sedentary adults in the United States and Europe. They were much more active, yet their daily total was not much higher.
2016: five populations, several hundred adults
A single small population could be an exception, so the next test was broader. In a 2016 paper, Pontzer and co-authors combined doubly labeled water measurements with accelerometer data from several hundred adults in five countries spanning a wide range of lifestyles. Expenditure rose with activity at the low end of the range, but among the more active participants the curve flattened: more movement was associated with little further increase in total daily energy spent. That pattern — rising, then levelling off — is the core of the constrained energy expenditure model.
2021: the human life course
The largest dataset came in 2021, when Pontzer and an international consortium published an analysis in Science of doubly labeled water measurements from more than 6,400 people, from newborns to people in their 90s. After adjusting for body size and composition, energy expenditure followed four phases: it climbs steeply in the first year of life, declines slowly through childhood and adolescence, holds remarkably steady from about age 20 to about 60, and falls after 60. The paper was not about exercise, but it reinforced the larger point that expenditure behaves like a regulated quantity rather than a simple sum of activities.
2021: measuring the compensation
The same year, a team led by Vincent Careau, with Pontzer among the co-authors, used that pooled database to ask how much of the energy spent on activity is offset by lower basal expenditure. Their analysis found that compensation does occur on average — part of the calories burned in activity is clawed back — and that the offset was larger in people with more body fat. Metabolic-ward research from Kevin Hall’s group at the U.S. National Institutes of Health has, separately, documented how the body adapts its expenditure after weight loss.
What changed in the model
The textbook picture was additive. Start with basal metabolic rate, the cost of staying alive; add the thermic effect of food, roughly a tenth of intake; then add activity. Every extra kilometre walked adds its calories to the daily total, in a straight line.
The constrained model replaces that line with a curve. Activity still costs energy, but as it becomes habitual the body appears to spend less elsewhere — candidates Pontzer discusses include resting processes, immune activity, stress responses and reproductive investment — so the total stays within a narrower band than the additive model predicts. In his 2021 book Burn, Pontzer describes the body as an adaptive system with a budget rather than a furnace that burns whatever it is fed.
Misreadings, checked against the data
| What people say | What the research supports |
|---|---|
| ”Exercise doesn’t burn calories.” | It does. The acute cost of activity is real; the constraint appears over weeks and months as spending elsewhere adjusts. |
| ”Exercise is useless for weight.” | Not supported. Activity is consistently associated with better weight maintenance after loss and has large benefits for heart, metabolic and mental health that do not depend on calorie burn. |
| ”So calories in, calories out is wrong.” | No. Energy balance still holds. What changed is how flexibly the “out” side responds when activity changes. |
| ”Your metabolism crashes in middle age.” | The 2021 life-course data show size-adjusted expenditure is stable from roughly 20 to 60. |
| ”Everyone compensates the same amount.” | Compensation varies between people, and in the 2021 analysis it was larger in people with more body fat. That variation helps explain why exercise programs produce such different results. |
What it means if you count calories
Calculated targets are hypotheses. Equations such as Mifflin-St Jeor or Harris-Benedict, multiplied by an activity factor, assume the additive model. For people who are habitually very active, that assumption tends to overstate expenditure, so a target built on it can be set too high. Our piece on why food calorie charts mislead covers the intake side of the same problem.
Your weight trend is the better measurement. If you log intake consistently and weigh yourself regularly, the relationship between the two over several weeks reveals your real expenditure, compensation included. That is the logic behind adaptive calorie targets in apps such as MacroFactor, which back-calculates expenditure from your weight trajectory, and PlateLens, whose adaptive target learns your expenditure from logged intake and weight trend and is part of its free plan. Either way, the target follows the body rather than an equation. If your log is inconsistent, though, no algorithm can recover the signal — the most common errors are listed in the calorie-tracking mistakes that quietly stall progress.
A plateau is information. When weight stops moving after months of steady intake, the usual explanation is that expenditure has adjusted and the gap has closed, not that effort has failed. The options are to accept the new level as maintenance or to lower intake modestly and watch the trend again.
Exercise numbers on screens are gross, not net. A treadmill that reports 480 calories is estimating the session’s cost. Over a week, the net change in your total may be smaller once compensation is counted. That is a reason to be careful about eating those calories back, not a reason to stop exercising.
A practical protocol
- Begin with an estimate. Use a standard equation to set a first target, and treat it as provisional.
- Collect four to eight weeks of data. Log intake as consistently as you can and track a smoothed weight trend rather than single weigh-ins.
- Adjust from the trend. If weight moves faster or slower than planned, change intake in small steps.
- Let intake do most of the work for weight change. The compensation literature points to the intake side as the more reliable lever.
- Keep training for its own sake. Strength work also matters for preserving lean mass while losing weight; our review of the 1 g per pound protein rule covers the intake that supports it.
People using structured eating patterns can apply the same approach; the evidence on one of them is reviewed in intermittent fasting after the hype cycle.
The bottom line
Pontzer’s work showed that daily energy expenditure is regulated: it rises with activity up to a point, then levels off as the body economizes elsewhere, and once size is accounted for it stays steady through most of adult life. It did not show that exercise fails to burn calories, that energy balance is wrong, or that movement is pointless. For anyone tracking food, the useful lesson is to treat calculated targets as a first guess, let a consistent log and a weight trend reveal the real number, and keep exercising for the benefits that never depended on the calorie display. More evidence reviews are collected in our nutrition science section.