Food and Behaviour Research

Donate Log In

Children’s Food and Mood: What Works, What Matters - BOOK HERE

Ultraprocessed Food: Addictive, Toxic, and Ready for Regulation

Robert H. Lustig  (2020) Nutrients 2020 12(11) doi:10.3390/nu12113401 

Web URL: Read this and related articles on MDPI


Past public health crises (e.g., tobacco, alcohol, opioids, cholera, human immunodeficiency virus (HIV), lead, pollution, venereal disease, even coronavirus (COVID-19) have been met with interventions targeted both at the individual and all of society. While the healthcare community is very aware that the global pandemic of non-communicable diseases (NCDs) has its origins in our Western ultraprocessed food diet, society has been slow to initiate any interventions other than public education, which has been ineffective, in part due to food industry interference. This article provides the rationale for such public health interventions, by compiling the evidence that added sugar, and by proxy the ultraprocessed food category, meets the four criteria set by the public health community as necessary and sufficient for regulation—abuse, toxicity, ubiquity, and externalities (How does your consumption affect me?). To their credit, some countries have recently heeded this science and have instituted sugar taxation policies to help ameliorate NCDs within their borders. This article also supplies scientific counters to food industry talking points, and sample intervention strategies, in order to guide both scientists and policy makers in instituting further appropriate public health measures to quell this pandemic.

1. Introduction: Pandemics and Public Health

We are in the midst of two pandemics. The COVID-19 pandemic had an identifiable start in January 2020. Yet despite media attention and warnings from scientists, many countries are experiencing a “second wave”; here in the United States, we never even cleared the first wave. There is no cure, at least not yet; all we have to mitigate this pandemic are public health efforts—social distancing, handwashing, and face masks—which do not seem to work very well voluntarily, unless made mandatory by authorities. The second pandemic, of non-communicable diseases (NCDs; type 2 diabetes, cardiovascular disease, fatty liver disease, hypertension, heart disease, stroke, cancer, and dementia), has been more insidious, slowly building over a 50-year time frame [1]. There is also no cure for this pandemic; all we have are educational efforts such as voluntary “diet and exercise”, which do not seem to work very well either.

NCDs now account for 72% of deaths [2] and 75% of health care dollars in the United States [3] and globally [2]; and the morbidity, mortality, and economic costs continue to climb. In the U.S., Medicare is expected to be insolvent by 2026, and Social Security will be broke by 2034 [4], due to both the loss of economic productivity combined with increased healthcare expenditures. Without young and healthy people paying into the system, old and infirm people cannot take out. The cost of these diseases is not limited to the U.S. [5], and NCDs have been declared a global health crisis by the United Nations (U.N.) [6]. Thus, NCDs pose an existential threat to the survival of each country, and indeed the entire planet. Identifying the cause(s) of NCDs, and upstream policy initiatives to mitigate them is of paramount importance.

Nonetheless, the world has recently faced down two other chronic disease pandemics, tobacco and ethanol; both caused by hedonic substances readily available for purchase, and both responsive to public health regulatory interventions. It was not until the U.S.’s Master Settlement Agreement and the World Health Organization (WHO) Framework Convention on Tobacco Control that we saw a reduction in cigarette consumption and reduction in lung cancer [7]. For alcohol, individual countries have passed their own public health ethanol regulatory efforts, with clear improvements [8].

2. Criteria for Public Health Regulation

The question for public health officials is whether there is something specific and identifiable that could be regulated on a global scale that could help to mitigate the pandemic of NCDs. While some behaviors can be mandated (e.g., mask-wearing), most are left up to each individual (e.g., exercise). Rather, targeting a substance or class of causative substances would be more effective, as predicted by the Iron Law of Public Health, which states that reducing availability of a substance reduces consumption, which reduces health harms [9]. Public health officials have identified the four criteria which must be met in order to be considered for public health regulation [10]:
  • Abuse (why can’t you stop?)
  • Toxicity (why do you get sick?)
  • Ubiquity (why can’t you escape it?)
  • Externalities (why does your consumption harm me?)
To generate enthusiasm for any public health regulatory effort, the science and the logic of each of these criteria must be obvious and inescapable. The goal of this treatise is to provide the science that ultraprocessed food in general, and sugar in particular, meet all four criteria, and should be considered as targets for regulation of the NCD pandemic by the public health community and by policymakers.
However, first we must deal with the “elephant in the room”; the mythology that calories are the cause of obesity, and obesity is the cause of NCDs. If this were the case, then the processed food industry can use the mantra that “any calorie can be part of a balanced diet”, and thus deflect criticisms of their products. In order to provide evidence for the specific roles of sugar and ultraprocessed food in the pandemic of NCDs, we must first confront and dispel this mythology, by demonstrating that obesity is not a cause of NCDs because normal-weight individuals get NCDs as well. We must also demonstrate that the effects of sugar and ultraprocessed food on NCD prevalence and severity are exclusive of inherent calories, and independent of effects on obesity [11].

3. Obesity Is a ‘Marker’, Not a Cause of Non-Communicable Diseases (NCDs)

Most clinicians mistakenly attribute the growing rise of NCDs to growing prevalence of obesity because of the quantity of the food ingested. This is untrue, for five separate reasons. (a) While obesity prevalence and diabetes prevalence correlate, they are not concordant [12]. There are countries that are obese without being diabetic (such as Iceland, Mongolia, and Micronesia), and there are countries that are diabetic without being obese, such as India, Pakistan, and China (they manifest a 12% diabetes rate). This is further elaborated looking at years of life lost from diabetes vs. obesity [13]. (b) Twenty percent of individuals with obesity are metabolically healthy and have normal life spans [14,15,16], while up to 40% of normal weight adults harbor metabolic perturbations similar to those in obesity, including type 2 diabetes mellitus (T2DM), dyslipidemia, non-alcoholic fatty liver disease (NAFLD), and cardiovascular disease (CVD) [17,18]. Indeed, in the U.S. 88% of adults exhibit metabolic dysfunction [19], while only 65% are overweight or obese—some normal weight people are metabolically ill as well. (c) The “Little Women of Loja” are a founder-effect cohort in Ecuador who are growth hormone-receptor deficient, and who become markedly obese yet are protected from chronic metabolic disease such as diabetes and heart disease [20]. (d) The secular trend of diabetes in the U.S. from 1988 to 2012 has demonstrated a 25% increase in prevalence in both the obese and the normal weight population [21]. (e) The aging process does not explain T2DM, as children as young as the first decade now manifest these same biochemical processes [22,23]. Now children get two diseases that were never seen before in this age group—T2DM and fatty liver disease. These two diseases used to be prevalent only in the elderly, or in those who abused ethanol.

These five lines of reasoning argue that obesity is a “marker” for the pathophysiology of NCDs (e.g., insulin resistance), but not a primary cause—because a percentage of normal weight people get NCDs as well, while a percentage of people with obesity are metabolically healthy. If obesity was a cause of NCDs, then one could by extension make the case that “eating is addictive”—but clearly neither are true. That young and normal weight people can contract these diseases suggests an exposure, rather than a behavior, at the root of the NCD pandemic, and that the quantity of the food is not the cause.

4. Ultraprocessed Food Is the Cause of NCDs

Rather, the quality of the food is the cause. Ultraprocessed food, defined as industrial formulations typically with 5 or more ingredients [24], is the category of food that drives NCDs [25], such as obesity [26,27], diabetes [28], heart disease [29], and cancer [30]. In particular, added sugar (i.e., any fructose-containing sweetener; sucrose, high-fructose corn syrup, maple syrup, honey, agave) is the prevalent, insidious, and egregious component of ultraprocessed food that drives that risk.

In this article, using scientific and legal evidence, I will elaborate three related arguments. First, I will demonstrate that ultraprocessed food is addictive because of the sugar that is added to it, and that the food industry specifically adds sugar because of its addictive properties. Second, I will highlight the specific mechanisms by which sugar is toxic to the liver, which leads to NCDs. Lastly, I will argue that added sugar is more appropriately defined as a food additive rather than as a food. In so doing, I will argue that added sugar, and by extension the entire ultraprocessed food category, meets these criteria established by the public health community for regulation of a substance (abuse, toxicity, ubiquity, externalities) [9].

5. Added Sugar Is Abused

The seminal role of the Western Diet in the pandemic of NCDs is unchallenged [31]. For instance, ultraprocessed food consumption correlates with body mass index (BMI) in the U.S. [26] and in 19 European countries [27]. As market deregulation policies of the 1990s took hold, fast food sales increased incrementally in all countries and cultures to which it has been introduced, along with commensurate increases in BMI [32]. Indeed, every country that has adopted the Western diet is burdened with the development of NCDs and their resultant costs [33].

However, the food industry continues to promulgate the argument that it is the quantity, not the quality of the foods that are to blame. This is not a semantic argument. Quantity is determined by the end user, a personal responsibility issue; while quality is determined by manufacturers, a public health issue. But what if the quality altered the quantity? Those that favored either view over the other would thus appear to be justified within their own stance. Indeed, this debate seems to have drawn to an academic stalemate [34,35,36]. This must be answered before any form of societal intervention can be contemplated.

5.1. ‘Food Addiction’ versus ‘Eating Addiction’

Recent revelations in the popular literature have alluded to the addictiveness of the Western diet [37,38], driving excessive consumption. Physiologic [39,40] and neuroanatomic [41] overlap between obesity and addiction pathways have been elucidated. Some investigators have argued that specific components of processed food, and in particular those in “fast food”, are addictive in a manner similar to cocaine and heroin [42,43]. The Yale Food Addiction Scale (YFAS) logs specific foods as having addictive properties [44], and a children’s YFAS also reveals that food addiction is common, especially in obese youth [45].

However, not everyone subscribes to this expanded view of specific foods having addicting properties. For instance, a group of academics in Europe called NeuroFAST does not accept the concept of food addiction, rather calling it “eating addiction” [46]. This group has proffered its own “eating addiction scale” in which all foods are treated similarly [47], and it is the behavior that distinguishes the phenomenon. These investigators state that even though specific foods can generate a reward signal, they cannot be addicting because they are essential to survival. In their own words:

“In humans, there is no evidence that a specific food, food ingredient or food additive causes a substance-based type of addiction (the only currently known exception is caffeine which via specific mechanisms can potentially be addictive). Within this context we specifically point out that we do not consider alcoholic beverages as food, despite the fact that one gram of ethanol has an energy density of 7 kcal [48]”.

NeuroFAST recognizes caffeine as addictive, but gives it a pass. Xanthine alkaloids are present naturally in many foods, yet caffeine is classified by the U.S. Food and Drug Administration (FDA) as a food additive. It is also a drug; we give it to premature newborns with underdeveloped nervous systems to stimulate the central nervous system (CNS) to prevent apnea. NeuroFAST also recognizes ethanol as addictive, and also gives it a pass. Natural yeasts constantly ferment fruit while still on the vine or tree, causing it to ripen [49], yet NeuroFAST acknowledges that purified ethanol is not a food. Rather, ethanol is a drug; we used to give it to pregnant women to stop premature labor.

Recently, another European group with food industry ties assessed the effects of specific foodstuffs on “eating dependence” in a cohort of university students, using weight gain as the metric of food addiction. In their study, they found no difference between fats and sugars as cause for weight gain [50]. However, as stated earlier, using weight gain as the metric of food addiction is inherently flawed.

In order to assess mechanism of effects of food on the addiction pathway in the brain, our group at UCSF studied a cohort of postmenopausal women with obesity who received orally the mu-opioid receptor antagonist naltrexone as a probe of the brain’s reward system. We found that the amplitude of cortisol responses and nausea generation in response to naltrexone correlated with symptoms of craving for sweet palatable foods in these women. These data suggest that naltrexone interfered with endogenous opioid peptide (EOP) tone that mediated these cravings. In so doing, we have discerned a phenomenon of “Reward Eating Drive” (RED), which belies those individuals with obesity who appear to respond excessively to hedonic food cues [51,52,53], and which is tied to the opioidergic component of the reward system in the brain, which is driven by sweet foods. Furthermore, using functional magnetic resonance imaging (fMRI) studies, other investigators have defined the prefrontal cortex as responsible for the response of sweet tastes as being “attractive” or “unattractive” [54].

5.2. Addictive Potential of Food Components

If there was a class of consumables that was uniquely addictive, it would have to be “fast food”. But is it just the calories, or is there something specific about fast food that generates an addictive response? Fast food contains four components whose hedonic properties have been examined: salt, fat, caffeine, and sugar [37,42].

5.2.1. Salt

In humans, salt intake has traditionally been conceived as a learned preference [55] rather than as an addiction. The preference for salty foods is likely learned early in life. Four- to six-month-old infants establish a salt preference based on the sodium content of breast milk, water used to mix formula, and diet [56]. Because energy-dense fast foods are relatively high in salt [57], in part as a preservative to reduce depreciation, the preference for salty foods is associated with higher calorie intake. For example, a study in Korean teens showed a correlation between frequent fast food intake and preference for saltier versions of traditional foods [58]. Another study examined 27 subjects undergoing opiate (mostly oxycodone) withdrawal and showed significant increases in fast food intake and weight gain over 60 days [59], suggesting “addiction transfer”. On the other hand, studies show that people can ‘reset’ their preference for less salty items. This has been demonstrated in adolescents deprived of salty pizza on their school lunch menu, and hypertensive adults who were retrained to consume a lower sodium diet over 8 to 12 weeks [55]. Furthermore, at low levels, salt intake is well known to be tightly regulated. For example, patients with salt-losing congenital adrenal hyperplasia who lack the mineralocorticoid aldosterone modulate have an obligatory salt loss, which modulates their salt intake [60], until appropriate doses of fludrocortisone are supplemented. The notion that human sodium intake is “physiologically fixed” had been used to criticize recent public health efforts to reduce sodium intake so drastically [61].

Nonetheless, the U.K. government engaged in a secret mass campaign to reduce public salt consumption by 30%, and saw a 40% reduction in hypertension and stroke without signs of withdrawal [62].

5.2.2. Fat

The high fat content of fast food is vital to its rewarding properties. Indeed, there may be a “high-fat phenotype” among human subjects, characterized by a preference for high-fat foods and weak satiety in response to them, which acts as a risk factor for obesity [63]. However, so-called “high-fat foods” preferred by people are almost always also high in carbohydrate (e.g., potato chips, pizza, or cookies). Indeed, adding sugar significantly enhances preference for high-fat foods among normal weight human subjects; yet there was no limit for preference with increasing fat content [64]. Thus, the synergy of high fat along with high sugar is likely to be more effective at stimulating addictive overeating than fat alone. However, these rewarding properties of fat appear to be strictly dependent on simultaneous ingestion of carbohydrate, as low-carbohydrate high-fat (LCHF) [65] and ketogenic diets [66] consistently result in reduced caloric intake, significant weight loss, and resolution of metabolic syndrome. In other words, fat increases the salience of fast food, but does not appear to be addictive in and of itself.

5.2.3. Caffeine

Caffeine is a “model drug” of dependence in humans [67], meeting the DSM-IV and DSM-5 criteria for tolerance, physiologic withdrawal, and psychological dependence in children [68], adolescents [69], and adults [70]. Headache [70], fatigue, and impaired task performance [68] have been demonstrated during withdrawal. While adolescents and children get their caffeine from soft drinks and hot chocolate, adults get most of their caffeine from coffee and tea [71]. These drinks average 239 calories and provide high amounts of sugar [72]. Soft drink manufacturers identify caffeine as a flavoring agent in their beverages, but only 8% of frequent soda drinkers can detect the difference in a blinded comparison of a caffeine-containing and caffeine-free cola [73]. Thus, the most likely function of the caffeine in soda is to increase the salience of an already highly rewarding (high sugar) beverage. These drinks may be acting as a gateway for caffeine-dependent customers to visit a fast food restaurant and purchase fast food [74].

5.2.4. Sugar

Other than caffeine, the component with the highest score on the YFAS is sugar [44]. Adding a soft drink to a fast food meal increases the sugar content 10-fold. Multivariate analysis of fast food transactions demonstrate that only soft drink intake is correlated with changes in BMI; not animal fat products [32]. While soda intake has been shown to be independently related to obesity and the diseases of metabolic syndrome [75,76], fast food eaters clearly consume more soft drinks. Sugar has been used for its analgesic effect in neonatal circumcision [77], suggesting a link between sugar and EOP tone. Indeed, anecdotal reports from self-identified food addicts describe sugar withdrawal as feeling “irritable”, “shaky”, “anxious” and “depressed” [78]; symptoms also seen in opiate withdrawal. Other studies demonstrate the use of sugar to treat psychological dependence [79]. Sugar craving can vary widely by age, menstrual cycle and time of day [80].

Sugar is added to food either as sucrose, high-fructose corn syrup (HFCS), honey, maple syrup, or agave. In general, each are assumed to consist of half fructose, half glucose; although this percentage has recently come into question when an analysis of store-bought sodas in Los Angeles revealed a fructose content as high as 65% [81]. This difference may be relevant, as fructose appears to generate a greater reward response and more toxicity than does glucose (see below).

5.3. Correlates of Addiction in Animals Exposed to Sucrose

In rodents, oral sucrose administration uniquely induces the acute reactant c-fos in the ventral tegmental area, implying activation of the reward pathway [82]. Furthermore, sucrose infusion directly into the nucleus accumbens reduces dopamine and µ-opioid receptors similar to morphine [83], and fMRI studies demonstrate the establishment of hard-wired pathways for craving [84]. Furthermore, sucrose administration to rodents induces behavioral alterations consistent with dependence; i.e., bingeing, withdrawal, craving, and cross-sensitization to other drugs of abuse [85]. Indeed, in one oft-quoted rat study, sweetness surpassed cocaine as reward [86].

5.4. Differential Effects of Fructose vs. Glucose vs. Fat on the Human Brain

Despite being calorically equivalent (4.1 kcal/gm), fructose and glucose are metabolized differently. Glucose is the energy of life. Glucose is so important that if you do not consume it, your liver makes it from amino acids and fatty acids (gluconeogenesis). Conversely fructose, while an energy source, is otherwise vestigial; there is no biochemical reaction in any eukaryote that requires it. Our research has shown that when provided in excess of the liver’s capacity to metabolize fructose via the tricarboxylic acid cycle, the rest is turned into liver fat, promoting insulin resistance, and resultant NCDs [87,88,89].

Physiologically, chronic fructose administration promotes fasting hyperinsulinemia and hypertriglyceridemia [90], which blocks leptin’s ability to cross the blood brain barrier [91], and attenuates leptin’s ability to extinguish mesolimbic dopamine signaling in rodents [92] and humans [93], thus promoting tolerance and withdrawal [94]. Furthermore, fructose does not suppress the stomach-derived hunger hormone ghrelin [95]. Through these pathways, fructose fosters overconsumption independent of energy need [96]. A comparison of the two monosaccharides demonstrates increased risk for bingeing with fructose (similar to sucrose) as opposed to glucose [97], suggesting the fructose molecule is the moiety that generates both reward and addiction responses.

Neuroanatomically, human fMRI studies show that acute glucose vs. fructose administration exert effects on different sites in the brain. One study infused each monosaccharide intravenously, and measured blood oxygenation level-dependent (BOLD) fMRI signal in cortical areas of the brain; glucose increased the BOLD signal in cortical executive control areas, whereas fructose suppressed the signal coming from those same areas [98]. Another study examined regional cerebral blood flow (rCBF) after oral glucose vs. fructose. With glucose, rCBF within the hypothalamus, thalamus, insula, anterior cingulate, and striatum (appetite and reward regions) was reduced, while fructose reduced rCBF in the thalamus, hippocampus, posterior cingulate cortex, fusiform, and visual cortex [99]. Consistent with other studies, fructose demonstrated lack of satiety or fullness in comparison to glucose. Furthermore, glucose increased “functional connectivity” of the caudate, putamen, precuneus, and lingual gyrus (basal ganglia) more than fructose; whereas fructose increased functional connectivity of the amygdala, hippocampus, parahippocampus, orbitofrontal cortex and precentral gyrus (limbic system) more than glucose [100]. In obese youth, the effects of oral fructose on dopamine activation of the nucleus accumbens is severely attenuated, suggesting down-regulation of dopamine receptors [101]. Lastly, the effects of fat and sugar both separately and together (adjusting for calories) on fMRI signaling have been assessed [102]. High-fat milkshakes increased brain activity in the caudate and oral somatosensory areas (postcentral gyrus, hippocampus, inferior frontal gyrus); while sugar increased activity in the insula extending into the putamen, the Rolandic operculum, and thalamus (gustatory regions). Furthermore, increasing sugar caused greater activity in those regions, but increasing fat content did not alter this activation. In other words, the fat increases the salience of the sugar, but it is the sugar that effectively recruits reward and gustatory circuits.

To summarize, added sugar (and specifically the fructose moiety) is unique in activating reward circuitry; fructose works both directly and indirectly to increase consumption; and that both obesity and chronic fructose exposure down-regulate dopamine receptors, requiring greater and greater stimuli to enact a reward-signaling effect (tolerance), a primary component of addiction.

5.5. ‘Food’ Addiction Is Really ‘Food Additive’ Addiction, and ‘Added Sugar’ Is a Food Additive

In the past, the concept of food addiction was not embraced by psychiatrists. For instance, the DSM-IV published in 1993 listed “substance use disorder” as requiring both tolerance and withdrawal as necessary criteria for the definition of addiction, and (apart from caffeine and ethanol) no foodstuff elicited withdrawal. However, as the public health difficulties stemming from addiction expanded, the definition, of necessity, expanded. The DSM-5 published in 2013 reclassified the field so as to include “behavioral addictions”, such as gambling (internet gaming was included in the Appendix as “requiring further study”). Thus, a revised set of criteria related to psychological dependence was proffered [103], including:
  • Craving or a strong desire to use;
  • Recurrent use resulting in a failure to fulfill major role obligations (work, school, home);
  • Recurrent use in physically hazardous situations (e.g., driving);
  • Use despite social or interpersonal problems caused or exacerbated by use;
  • Taking the substance or engaging in the behavior in larger amounts or over a longer period than intended;
  • Attempts to quit or cut down;
  • Time spent seeking or recovering from use;
  • Interference with life activities;
  • Use despite negative consequences.
However, food addiction was not codified in the DSM-5. Nonetheless, systematic reviews of the literature demonstrate that ultraprocessed foods have the highest addictive potential due to their added sugar content [104]. While sugar itself does not exhibit the DSM-IV criteria of classic tolerance and withdrawal, sugar clearly meets the DSM-5 requirements of tolerance and dependence (use despite conscious knowledge and recognition of their detriment).

Coca leaves are medicinal in Bolivia, yet cocaine is a drug, and regulated. Opium poppies are also medicinal, but morphine is a drug, and regulated. Caffeine is found in coffee (medicinal for many), yet concentr


For further information on this topic please see: