chatGPT:
Paper summary
This review, “Physical activity and metabolic rates in humans”, examines how physical activity and exercise affect whole-organism energy metabolism, especially the relationship between:
- AEE: active energy expenditure
- REE: resting energy expenditure
- DIT: diet-induced thermogenesis
- TEE: total energy expenditure
The central equation is:
TEE = REE + AEE + DIT
The paper compares three models:
-
Additive model
Increased physical activity simply adds to total energy expenditure. If AEE rises, TEE rises by roughly the same amount, while REE is unchanged. -
Stress model
Exercise increases AEE and also transiently raises REE through excess post-exercise oxygen consumption (EPOC), reflecting repair, recovery, mitochondrial biogenesis, immune activation, tissue remodeling and other adaptive processes. -
Constrained energy model
Increased AEE is offset by reduced REE, so TEE remains relatively constant. This model implies metabolic compensation: the body reduces energy spending on other functions when activity rises.
The authors argue that much support for the constrained energy model comes from problematic analyses, especially where AEE is calculated from TEE and REE, creating mathematical coupling and spurious negative correlations. They conclude that the strongest available evidence supports the additive model, with some support for a short-term stress/EPOC effect, and much weaker support for long-term metabolic constraint.
The paper then connects these metabolic models to health and ageing. It argues that physical activity improves healthspan partly by diverting energy away from processes that may improve short-term reproductive success but harm long-term health, such as excess visceral fat storage and high reproductive hormone exposure. It also argues that exercise-induced stress activates repair and maintenance pathways: antioxidant production, DNA repair, autophagy, heat-shock proteins, anti-inflammatory signalling, angiogenesis, mitochondrial biogenesis, muscle and bone repair, arterial remodeling and cardiac adaptation.
A major evolutionary framing is that humans evolved to be unusually physically active compared with other apes. Modern inactivity is therefore presented as an evolutionary mismatch, because many repair and capacity-building systems may require regular physical activity for full activation.
Novelty
The novelty is not a new dataset but a conceptual and methodological reanalysis of a major debate in exercise metabolism.
The paper’s most novel contributions are:
First, it directly contrasts the additive, stress and constrained energy models in both longitudinal and cross-sectional terms. This is useful because the constrained energy model is often discussed using cross-sectional data, even though its strongest prediction is longitudinal: increasing a person’s AEE should reduce their REE over time.
Second, it gives a strong methodological critique of the constrained energy literature. The authors emphasize that if AEE is calculated as something like 0.9TEE − REE, then using that calculated AEE to test its relationship with REE risks building a negative relationship into the analysis. That is a central technical objection.
Third, the paper integrates evolutionary biology, life-history theory and exercise physiology. It does not just ask whether exercise burns calories; it asks how exercise reallocates energy among reproduction, storage, repair, maintenance and capacity-building.
Fourth, it highlights the possibility that humans may be unusual among mammals in deriving major healthspan benefits from sustained physical activity, because humans evolved as highly active, long-lived primates with substantial post-reproductive lifespan.
Critique
The paper is persuasive in its critique of simplistic versions of the constrained energy model, but it may somewhat overcorrect toward the additive model.
A key strength is its insistence that TEE, REE and AEE must be measured independently. Many claims about compensation are weakened if AEE is derived by subtracting REE from TEE. The paper is also right that PAL-type ratios and body-mass-adjusted TEE comparisons can be misleading.
However, there are several limitations.
First, the review may understate real compensation outside REE. Even if REE does not fall, people may compensate through reduced non-exercise activity, increased food intake, altered movement economy, or behavioural fatigue. The authors acknowledge this, but the health and obesity debate often depends on these broader compensations, not just REE.
Second, the constrained energy model may be more plausible in specific settings: chronic high activity, energy limitation, infection burden, reproductive stress, older age, or extreme endurance activity. The review treats the broad human evidence as mostly additive, but different ecological contexts may produce different compensation patterns.
Third, REE is a coarse endpoint. A fall in reproductive hormone production, immune activity, inflammation, sympathetic tone or tissue turnover might not produce a large detectable fall in whole-body REE, yet could still represent biologically meaningful reallocation.
Fourth, the evolutionary argument is plausible but partly speculative. The idea that exercise uniquely activates human anti-ageing repair systems is attractive, but the paper itself notes that direct comparative evidence is limited. Rodent exercise studies often show modest or inconsistent lifespan effects, which complicates the claim that activity robustly extends lifespan across mammals.
Fifth, the review focuses heavily on energy allocation, but many benefits of exercise are not easily reducible to calorie flows: mechanical loading, shear stress, lactate signalling, myokines, mitochondrial quality control, glucose partitioning, immune training and neurotrophic signalling may matter as much as gross energy expenditure.
Overall assessment
This is a strong review that challenges the popular claim that total energy expenditure is tightly constrained in humans. Its most valuable contribution is methodological: it shows that some evidence for metabolic compensation may arise from circular calculations and regression artefacts.
The paper’s conclusion is reasonable: in most available human data, increased physical activity appears more additive than constrained, with short-term stress-related increases in REE through EPOC and repair processes. But the broader question of compensation remains open, especially when behavioural compensation, food intake, ecological stress and long-term adaptation are included.