The Epistemology of Human Nutrition: How Do We Know What Humans "Should" Eat?

Human nutrition as a field faces a fundamental epistemological question: How can we reliably determine what constitutes a species-specific, species-appropriate diet for Homo sapiens? Two broad paradigms compete for primacy in answering this.

The evolutionary / anthropological framework assumes humans, like all species, have a biologically adapted diet shaped by millions of years of natural selection. We can reconstruct this "species-appropriate" template using robust, historical sciences.
The modern epidemiologic framework treats human nutrition as largely a post-agricultural (or even post-industrial) problem. It relies on contemporary observational studies, RCTs, and meta-analyses — often referred to as the "pyramid of evidence" in evidence-based medicine — to derive dietary recommendations.

This page contrasts these approaches, highlighting their respective strengths, limitations, and implications for what we consider reliable knowledge in nutrition.

Approach 1: Assuming a Species-Specific, Evolutionarily Appropriate Diet

This perspective starts from first principles of biology: every species has dietary adaptations shaped by evolutionary pressures. For humans, the relevant timeframe spans ~2.6 million years of genus Homo (with deeper roots in hominin evolution), during which natural selection operated under conditions very different from the last ~10,000 years of agriculture.

Key Supporting Disciplines and Methods

These fields provide convergent, relatively "hard" evidence from the deep past:

Paleoanthropology & archaeology — Fossil records, tool use, and site evidence show shifts toward increased meat/marrow consumption ~2.6 Ma, brain expansion in Homo erectus, fire control, and broad-spectrum foraging.
Stable isotope analysis (δ¹³C and δ¹⁵N in bone collagen) — Reconstructs dietary protein sources. Upper Paleolithic humans often show high animal protein intake, including significant aquatic resources in many populations. Compound-specific amino acid analysis refines this, confirming freshwater fish or marine inputs in some Late Upper Paleolithic groups.
Paleopathology — Skeletal evidence of health/disease. Hunter-gatherer remains frequently show fewer signs of the "diseases of civilization" (e.g., caries, anemia from grain-heavy diets, metabolic conditions) compared to early agricultural populations.
Evolutionary theory — Mismatch hypothesis: post-agricultural diets (high refined carbohydrate, low nutrient density, novel foods like dairy/grains for non-adapted groups) create discordance with our genome, contributing to chronic diseases.

This framework views humans as having a flexible but bounded omnivorous adaptation — capable of wide variation (e.g., Inuit vs. Kitavan diets) yet optimized around animal-source foods for dense nutrition, combined with gathered plants, and minimal ultra-processed items.

Strengths include:

Grounded in immutable evolutionary processes.
Long timescales reduce short-term confounders.
Convergent lines of evidence from independent methods.

Approach 2: Relying on Modern Epidemiologic Science

This paradigm assumes nutrition science is largely a modern endeavor. Humans can eat almost anything in the short term, so optimal diets must be discovered empirically using tools of evidence-based medicine applied to contemporary populations.

Core Methods ("Pyramid of Evidence")

Observational studies (cohort, case-control) — Track associations between self-reported diet and disease outcomes.
Randomized controlled trials (RCTs) — Test specific interventions (e.g., low-fat vs. control).
Meta-analyses & systematic reviews — Pool results for summary estimates.

These feed into guidelines (e.g., food pyramids, plates, or national recommendations).

Major Criticisms and Limitations

Nutritional epidemiology faces well-documented challenges that undermine causal inference:

Confounding — Impossible to fully adjust for lifestyle, socioeconomic, or reverse causation factors (e.g., sick people change diets).
Measurement error — Self-reported intake (food frequency questionnaires) is notoriously inaccurate, especially over decades.
Short duration & compliance issues in RCTs — Dietary RCTs are hard to blind, long-term adherence is poor, and baseline status varies.
Heterogeneity & bias in meta-analyses — Publication bias, selective reporting, and inappropriate statistical models (e.g., choosing fixed vs. random effects post-hoc) distort results.
Mismatch with nutrient complexity — Diet is not a single drug; food synergies, dose-response, and individual variability are poorly captured.

Many reviews show observational findings often fail to replicate in large pragmatic RCTs, and certainty of evidence remains low for most diet-disease links.

Comparison and Implications

Aspect	Evolutionary Framework	Modern Epidemiologic Framework
Time scale	Millions of years	Decades (mostly post-1950s)
Primary evidence type	Historical / biological (isotopes, fossils)	Contemporary human studies (observational + RCT)
Causal strength	Indirect but evolutionarily grounded	Direct but heavily confounded
Main assumption	Humans are adapted to ancestral environments	Optimal diet discoverable via current methods
Robustness to confounders	High (deep time averages out noise)	Low (myriad modern biases)
Policy implication	Return toward ancestral patterns (more whole animal foods, fewer processed carbs)	Evidence-based tweaks (e.g., reduce saturated fat, increase fiber) — often shifting

The evolutionary approach offers a unifying paradigm rooted in biology, while epidemiology provides proximate associations that are frequently fragile and revised.

Many researchers argue the former provides the stronger epistemological foundation for a default species-appropriate diet, treating modern epidemiology as supplementary for fine-tuning within that template — especially given the explosion of chronic disease since the agricultural and industrial revolutions.

Approach 1: Assuming a Species-Specific, Evolutionarily Appropriate Diet​

Key Supporting Disciplines and Methods​

Approach 2: Relying on Modern Epidemiologic Science​

Core Methods ("Pyramid of Evidence")​

Major Criticisms and Limitations​

Comparison and Implications​

Further Reading​