Nucleus of the hyper-palatable-foods cluster. A hyper-palatable food (HPF) is a food whose
combination of palatability-inducing ingredients (fat, simple sugar, carbohydrate, sodium) is high enough
to plausibly override normal satiety and heighten food reward. The term was long used descriptively and
un-standardized (fast food, sweets) — Fazzino’s contribution is to replace that label with a
data-derived quantitative definition on nutrient composition, making palatability an admissible,
measurable exposure rather than a vibe. This page holds the construct; it does not yet hold an
outcome effect (see Evidence state — this is a definition paper, not an outcome study).
(inferred from Fazzino et al., 2019)
The definition — three nutrient clusters
A systematic review of the descriptive HPF literature (14 articles -> 75 nonduplicate food descriptors) was quantified on four ingredients in ASA24 nutrition software and clustered (two independent raters, data-visualization). Three clusters emerged, covering 93% (70/75) of the literature items:
| Cluster | Threshold combination |
|---|---|
| FSOD (fat + sodium) | > 25% kcal from fat AND >= 0.30% sodium by weight |
| FS (fat + simple sugar) | > 20% kcal from fat AND > 20% kcal from sugar |
| CSOD (carbohydrate + sodium) | > 40% kcal from carbohydrate AND >= 0.20% sodium by weight |
«Findings revealed that HPF fall into three distinct clusters based on similar levels of ingredients: (1) FSOD (> 25% kcal from fat and >= 0.30% sodium by weight), (2) FS (> 20% kcal from fat and > 20% kcal from sugar), and (3) CSOD (> 40% kcal from carbohydrates and >= 0.20% sodium by weight).» (Fazzino et al., 2019)
The load-bearing idea is the combination, not any single nutrient. Each cluster names two ingredients above threshold; no single-nutrient cluster exists. The mechanism is synergy: «foods designed to be hyper-palatable … may contain combinations of palatability-inducing ingredients (fat, sugar, carbohydrates, and/or sodium) at moderate to high levels that may circumvent physiological satiety mechanisms and activate brain reward neural circuitry.» (Fazzino et al., 2019) Theoretical basis is sensory-specific satiety (SSS) circumvention — two palatability ingredients delay eating cessation more than either alone. Precedent human evidence: «a combination of fat and sodium (FSOD) may synergistically enhance food palatability and increase consumption by up to 30% (16,17).» (Fazzino et al., 2019) This is a unit-of- analysis shift: from the single nutrient (the object of most guidance — sodium, free sugars, fat) to the co-occurrence of nutrients as the palatability driver. Single-nutrient targets, by construction, cannot see it. (inferred from Fazzino et al., 2019)
Prevalence — most of the US food supply
Applied to the USDA FNDDS (7,757 solid foods): «The HPF criteria identified 62% (4,795/7,757) of foods in the FNDDS that met criteria for at least one cluster.» (Fazzino et al., 2019) Of HPF items, «Most HPF items (70%; 3,351/4,795) met criteria for the FSOD cluster.» (Fazzino et al., 2019) (FS 25%, CSOD 16%; < 10% met more than one). Fat+sodium is thus the dominant hyper-palatable profile in the US supply, and HPF spans > 70% of meats, grains and eggs. Read as a shape, not a target: 62% is the prevalence of a newly-drawn line, and the cut points «should not be assumed to be fixed or final» (Fazzino et al., 2019) — the number is where the data-visualization drew the boundary (n=75 items, no advanced statistics), so treat it as a first draft of a definition, not a measured constant.
What HPF is NOT — distinct from ultra-processing and from energy density
This is the disambiguation (type-B) that earns the construct its place. HPF is a composition / reward construct; it is neither NOVA ultra-processing (a manufacturing-degree classification) nor energy density (kcal/g). Fazzino positions it against both: «research on HPF lags behind other related research with established dietary indices, such as energy density (28) and ultraprocessed foods (29).» (Fazzino et al., 2019)
- Not energy density. HPF energy density ranged < 1.0 to 8.4 kcal/g, and «Almost half of HPF items (49%; 2,337/4,809) had low energy density (< 2 kcal/g) (50).» (Fazzino et al., 2019) A food can be hyper-palatable and low-energy-density (vegetables cooked in fat/sodium; sweetened breakfast yogurt) — so the two indices come apart, and HPF is not a rename of energy density. -> Energy Density
- Not NOVA ultra-processing. UPF is defined by industrial manufacturing; HPF by nutrient composition — measurable directly from a nutrition label, not from a manufacturing narrative. They correlate but are different objects. The overlap magnitude across the US food supply is quantified in Overlap and 30-year trend below (Sutton 2023, same lab).
- More decision-usable than the NOVA category. Because HPF is composition-based, it decomposes into the fat/sugar/sodium/carb combinations a person can read off a label — sidestepping the Is the Food Category Doing Any Work complaint against NOVA (that a food’s HPF status turns on manufacturing detail a shopper cannot apply). Whether hyper-palatability itself does causal work beyond its component nutrients is the same open category question, unresolved here.
- Not Food Addiction either — composition vs behaviour
[type-B]. HPF is a nutrient-COMPOSITION definition, agnostic about behaviour; the food-addiction construct is a BEHAVIOURAL/reward construct (YFAS-measured loss of control), agnostic about exact composition. They overlap on the implicated foods (fat + refined carbohydrate) but a claim under one does not transfer to the other — a food being HPF by threshold does not make it “addictive”, and a high YFAS rating does not show an HPF profile. Schulte 2015 supplies the reward-side predictor evidence (processing/fat/glycemic-load predict addictive-like eating) that HPF’s composition definition leaves open. (inferred from Fazzino et al., 2019)
(inferred from Fazzino et al., 2019)
Overlap and 30-year trend — three constructs, most of the supply, but not interchangeable [type-F]
Sutton 2023 (same Fazzino lab — a construct extension, not independent-E) applies all three definitions to four USDA datasets spanning the US food supply 1988-2018 (>6000 solid foods each; beverages/SSBs excluded). It cashes what this page previously only flagged as awaited: the overlap magnitude and the temporal trend. (Sutton et al., 2023)
Prevalence and trend. «58% to 65% of foods were classified as UPF, 55% to 69% as HPF, and 37% to 47% as HED.» (Sutton et al., 2023) All three rose 1988->2018 (p<0.001), but at different rates: «HPF evidenced the largest increase (14%) and UPF evidenced the smallest (4%) over time» (HED +10%). UPF has «stabilized since 2006»; HPF climbed the most and most recently. This is a food-SUPPLY fact, not an exposure-response claim — it says the shelf has shifted toward hyper-palatable composition, not that any person’s outcome changed. The parallel to rising obesity rates is explicitly the paper’s future work, not its finding.
Overlap is high but incomplete, and asymmetric. «Moderate to high overlap in foods (40%-70%) across definitions.» (Sutton et al., 2023) Directionally (share of X also classified as Y, 2018): ~78% of UPF are also HPF; ~73% of HPF are also UPF; HED overlaps the most (up to ~86% of HED items are also HPF) and is the least distinct construct. «Approximately one third of foods… met criteria for all three definitions (UPF, HPF, and HED), indicating a substantial portion of the food supply may be particularly obesogenic.» (Sutton et al., 2023) Only a small minority (~15-20%) met none.
The distinctness is the decision-relevant payoff — each construct captures a food set the others miss. The disambiguation is not merely verbal: the uniquely-classified foods differ by mechanism.
- Distinctly HPF (not UPF, not HED): «HPF were primarily fresh or whole food items prepared with palatability-enhancing ingredients during cooking (primarily fat and salt).» (Sutton et al., 2023) HPF sees home-cooked whole foods that UPF (a manufacturing classification) cannot — the supply-scale vindication of the preparation-not-food-identity refinement below.
- Distinctly HED (not UPF, not HPF): «uniquely HED items had naturally high energy density (i.e., nuts), but did not have other added palatability-inducing nutrients and were raw.» (Sutton et al., 2023) Raw nuts and dried fruit are energy-dense but neither engineered nor hyper-palatable — showing why energy density alone is a poor proxy for the reward construct.
Consequence for reasoning (the transportability guard). Because overlap is high but far from complete, a claim established under one definition does not transfer to another — an outcome shown for UPF is not thereby shown for HPF or HED, and vice versa. Sutton’s own construct-choice rule: match the definition to the hypothesized mechanism — «UPF for endocrine-disrupting chemicals and additives, HPF for exaggerated palatability that may yield overconsumption, and HED, which provide more kilocalories per bite.» (Sutton et al., 2023) The three are correlated indices of an obesogenic supply, not interchangeable exposures.
(inferred from Sutton et al., 2023)
Two operational refinements
- Preparation, not food identity, is the determinant. Across 478 aggregated food types, 81% could be prepared to meet HPF, but among 43% (204/478) >= 70% of preparation methods did not — «the findings suggest that method of preparation, as opposed to food item, determined whether a food met HPF criteria.» (Fazzino et al., 2019) So HPF is a property of the dish as prepared, not of the raw ingredient — a chicken leg is HPF cooked in butter, not grilled. This is a substitution/implementation lever (Layer 3): the same food moves in and out of HPF by how it is cooked.
- Reduced/low labels do not exempt a food. Of items labeled reduced/no fat, sugar, salt or calorie, 49% still met HPF; «Most notably, 80% of items labeled as reduced fat or calorie (102/127) met criteria for HPF.» (Fazzino et al., 2019) — a weight-management marketing label is not evidence a food lacks a hyper-palatable profile.
Evidence state — a definition, not an outcome finding [insufficient evidence for outcomes]
This paper carries the construct’s admissibility, not its harm magnitude. It is a cross-sectional food-database analysis (tier moderate) with no patient-important outcome data. The definition earns HPF a place as a specified, falsifiable, measurable exposure (the telos operationalize before admitting test — passed); the claim that HPF causes overeating, weight gain, or disease is not established here and is held at insufficient evidence, not benefit/harm. Fazzino says so: the criteria «should be adapted, refined, and tested for their predictive utility» for overeating, weight gain, obesity risk and T2D. (Fazzino et al., 2019) Symmetric standards: a foods engineered to be addictive framing gets the same bar — the definition is admissible, the harm is not yet evidenced.
(inferred from Fazzino et al., 2019)
Gaps and open questions
- G-gap — the HPF -> patient-important-outcome SR/MA does not exist. There is no held (or, as of this ingest, known) systematic review linking HPF exposure to weight gain, obesity, T2D or mortality. Until one exists, HPF cannot enter the Layer 1 - Ranking Interventions for a Stratum hierarchy as a sized lever — only as a candidate mechanism. — an HPF-exposure -> hard-outcome pooled analysis.
- Live question — is hyper-palatability the mechanism behind the UPF overconsumption effect? Fazzino proposes HPF as a candidate for the unidentified mechanism in the Hall UPF-weight-gain trial: «a recent study established a causal relationship between ultraprocessed food consumption and weight gain; however, the authors noted that their study did not identify the mechanism … (52).» (Fazzino et al., 2019) But this is not a joined issue: Hall matched the UPF and unprocessed diets on composition and found the excess intake was not explained by pleasantness/appetite ratings (those did not differ) — running through eating rate, energy density and protein dilution instead -> Ultra-Processed Food and Health Outcomes. Fazzino asserts palatability may be the driver; Hall’s design could not test HPF-vs-non-HPF palatability. So the two do not yet contradict — they answer different questions. Recorded as an open question, not a filed tension. (inferred from Fazzino et al., 2019)
- G-gap — no dose or shape. HPF is a binary threshold membership, not a dose. Whether “more hyper-palatable” (further above threshold, or meeting more clusters) buys more overconsumption is unstudied — the definition is a line, not a curve.
Self-critique [run 2026-08-30, re-run after Sutton 2023 extension, before commit]
- Over-claim check: every strength claim is scoped to definition / prevalence / overlap / discriminant validity; the page does not assert HPF causes any outcome (the central risk for an engineered addictive foods construct). The Sutton trend (HPF +14%) is stated explicitly as a food-SUPPLY fact, NOT an exposure-response or harm claim, and the obesity-rate parallel is flagged as the paper’s future work. The insufficient-evidence-for-outcomes state is unchanged.
- Independence check (Sutton not E): Sutton 2023 shares senior author Fazzino with the definition
paper and explicitly re-applies the Fazzino 2019 [9] definition — so it is a type-F extension/refinement
of the same construct, NOT type-E independent backing. Marked F; no
[E-independent]token written;sources:carries both because Sutton contributes distinct extracted claims (overlap, trend, distinct-food sets) satisfying the dual test. - Not-joined check (Fazzino vs Hall): unchanged — the two answer different questions; open question, not a tension.
- Transportability guard added: the overlap being high-but-incomplete now carries an explicit non-transfer rule (a claim under one construct does not transfer to another), grounding the type-B disambiguation in a decision consequence rather than a merely verbal distinction.
- Confidence: stays
low— both sources are moderate-tier definitional/prevalence studies from one lab; still no outcome evidence, so the HPF->hard-outcome SR gap remains open ().