A lot of people assume exercise calories work like simple math: burn 300 extra calories and total daily energy expenditure should rise by 300. This paper suggests the body often compensates for added activity by reducing energy spent elsewhere, so the net increase in daily expenditure is smaller than expected.

Overview

• What did they test? Pontzer and Trexler reviewed evidence comparing two models of total daily energy expenditure (TEE): an additive model, where exercise calories stack on top of the rest of daily expenditure, and a constrained model, where increases in physical activity are partly offset by reductions in other energetic processes. They examined data from human exercise interventions in free-living human populations. 
• What did they find? In human aerobic exercise interventions, total daily energy expenditure increased by only about 30% of the amount predicted by additive models. Compensation tended to be greater with aerobic exercise, greater still when aerobic exercise was paired with diet restriction, and was much less clear in the small number of resistance training cohorts. They also found that reductions in basal metabolic rate and especially sleeping metabolic rate appear to contribute to compensation.
• What does it mean for you? Exercise still raises energy expenditure, but not by as much as calculators or cardio machines imply. This helps explain why exercise-induced weight loss is often smaller than expected. It also suggests that the body adapts to sustained increases in activity.

What’s the Problem?

Exercise is one of the most reliable ways to improve health, but its role in weight loss has always been more complicated. The assumption is that energy expenditure during exercise adds onto daily energy expenditure. Burn 300 calories during a run, and total daily energy expenditure should rise by about 300 calories. This is the additive model, and it is the logic behind many calorie calculators and exercise-based weight-loss predictions.

The constrained model makes a different argument. It suggests that when physical activity increases, the body may partially compensate by reducing energy spent elsewhere. That could include reductions in resting or sleeping metabolic rate, less spontaneous movement, greater movement efficiency, or lower energy allocation to other functions like the endocrine or immune system. In this model, a 300-calorie run may still increase daily energy expenditure, but the net increase may be much smaller than 300 calories.

This debate has become more nuanced over time. Pontzer’s earlier work helped popularize the idea that total energy expenditure may be constrained, with ecological and intervention data suggesting that highly active people do not always have proportionally higher daily energy expenditure 1 2. However, other studies have challenged this model. For example, Willis et al. found that physical activity and TEE were generally consistent with an additive pattern in older U.S. adults, but the relationship looked more constrained among people in negative energy balance 3. 

This is where the debate gets tricky. Both models can predict that more active people will have higher total energy expenditure. A positive association between activity and TEE does not automatically prove the additive model, and a weak association does not automatically prove the constrained model. On top of that, measurement error, regression dilution, differences in devices, variation in body size, and changes in energy intake can all blur the relationship. Reviews of the weight-loss literature generally agree that exercise can produce weight loss, but usually less than predicted from the prescribed exercise energy expenditure 4. In other words, exercise works, but it’s not as clean as we once thought.

Purpose

The purpose of this paper was to develop a quantitative framework for testing additive versus constrained models of total energy expenditure and then compare those predictions against data from human studies.

Hypothesis

The authors did not frame it as a traditional hypothesis-testing study, but the paper evaluates whether available evidence better supports a constrained or additive model of TEE.