Ultra-Processed Evidence

3 Jan 2025 | Critical Thinking

Chris van Tulleken’s “Ultra-Processed People: Why Do We All Eat Stuff That Isn’t Food … And Why Can’t We Stop?”, on the dangers of Ultra-Processed Food (UPF), was one of the most successful books of 2023. It was a #1 Sunday Times Bestseller and an Economist, Daily Mail, Good Housekeeping, and Amazon Book of the Year. Even now (January 2025) it’s still #9 in the Amazon charts, having spent 88 weeks in the top 20.

Due to its popularity, I’d heard about it before reading it, and the punchlines resonated strongly with me. I try to eat home-cooked meals wherever possible, consume very little “junk” food, and are sceptical about the companies that produce them. Some of them – such as Ben & Jerry’s, Tony’s Chocolonely, and Mars – I come across in my day job as they’re heralded as purposeful businesses, which I find surprising given their contribution to global obesity.

But a couple of the conclusions that I’d heard gave me pause. First, it made strong black-and-white statements, claiming that all UPF is bad, in all quantities and for all people. This seemed to exemplify the black-and-white thinking highlighted in Chapter 2 of May Contain Lies. Second, it made strong causal claims that UPF causes obesity. But in Chapter 6 I explain how it’s very difficult to show that diet causes anything, because it’s correlated with a whole host of other factors. Even though infant formula (a prime UPF according to van Tulleken’s definition) is correlated with worse child health, it’s not causal – family characteristics (such as poverty) lead parents to bottle-feed, and it’s those characteristics that cause ill health, not the formula itself.

Yet those were only the conclusions that I’d heard, and perhaps people were paraphrasing them inaccurately. I want to read the book itself, to see what it actually claimed and how strong the evidence was. And not for an intellectual exercise, to see if I could catch the author out, but because evidence matters. The claim that bottlefeeding harms infant health – one that van Tulleken also makes in his book – imposes significant shame, pressure, and stress on mothers who are unable to or choose not to breastfeed, but the evidence is far weaker than often portrayed. Similarly, if UPFs aren’t as bad as van Tulleken alleges, then people who don’t have the time to home-cook every meal (or money to hire someone to do so) aren’t damaging their health. Emily Oster’s excellent series of books on parenting used the statistical rigour of an economist to re-examine many commonly-held views on parenting. By debunking several myths, her work allows for better, more relaxed parenting. I wanted to apply the same statistical rigour to this book. Should we really beat ourselves up if we treat ourselves to a Ben & Jerry’s ice cream?

What The Book Claims

Let me first stand in the author’s corner. The book draws the following conclusions which, if true, are very important:

Food’s contribution to obesity depends far less on its fat and sugar content than whether it’s UPF.
Profit-seeking food companies design UPFs to be maximally addictive. This addictiveness is so strong that it outweighs our body’s natural satiation response, so willpower won’t stop us eating it.
Exercise doesn’t help. The calorific content of UPF is so high that exercise will barely make a dent, and a workout leads to you taking it easy the rest of the day anyway.

It’s easy to see why the book has become popular: these conclusions feed on our confirmation bias. Most people mistrust giant corporations and view them as extracting profit at the expense of society. The claims also allow us to blame obesity on others – it’s not due to our own lack of willpower or failure to exercise, but those evil companies. As the book’s back cover states: “It’s not you, it’s the food”, and as the introduction reassures us “If you’re living with obesity, it isn’t due to a lack of willpower; it isn’t your fault.”

What I Like About The Book

The appeal of the conclusions doesn’t mean they’re wrong: sometimes what we want to be true is actually true. And there are several things I like about the book. The main one is the importance of looking beyond traditional nutritional labels. The nutritional information on a food product lists its total fat, saturated fat, salt, and sugar content; in the UK, red, amber, and green “traffic light” labels tell customers whether these are high, medium, or low. But, just as my own work finds evidence of financial box-ticking, van Tulleken highlights nutritional box-ticking: reducing sugar content but replacing it with potentially harmful artificial sweetener. What’s not in the labels may matter even more than what’s in them. Plus, all nutrients aren’t created the same; if something is highly processed, it may be easier to eat in large quantities.

This takeaway isn’t hugely surprising. Aged 7, I went to a school that provided no school meals so we had to bring packed lunches. At lunchtimes we’d point out if our food had artificial preservatives or colourings (in addition to chatting about more normal things, such as trading Garbage Pail Kids stickers). We know that fresh vegetables are different from canned. At times, I felt the book was knocking down a straw man: if you only looked at fat, salt, and sugar, it wouldn’t matter if food was mouldy. But even if we knew this subconsciously, it’s good for it to be emphasised explicitly, and the book does so in a persuasive and readable way.

I also like how the book is sceptical about evidence. Rather than ignoring evidence against the book’s conclusions, it acknowledges it up-front and highlights its limitations. It also diligently uncovers the financial ties that the authors of these studies had to the food industry – in many cases, ties that the authors failed to disclose themselves. As Chair of the Ethics Committee of the Financial Management Association (one of my profession’s associations) I initially appreciated this, although I later learned that some ties were wrong (they were established several years after the research was conducted) and others were significantly exaggerated (being reimbursed travel expenses). Van Tulleken also appears to recognise challenges in his own evidence (is it UPF that causes obesity, or do poorer families tend to eat UPF and it’s poverty that causes obesity?) and to address them, yet as I’ll soon discuss, he only does so selectively.

But are the book’s conclusions as strong as the author – and the long list of celebrity endorsers and influential newspaper reviews – claim? Unfortunately, there are several important flaws.

The Definitions of UPF are Vague and Inconsistent

For a book about UPF, it is hopelessly vague about defining what a UPF is, and chops and changes its definition throughout. The back of the book says that “There’s a long, formal scientific definition, but it can be boiled down to this: if it’s wrapped in plastic and has at least one ingredient that you wouldn’t find in your kitchen, it’s UPF.” This is also the definition given in the book’s introduction.

But when I got into the book itself, the first definition comes in Chapter 2, where the author describes the NOVA classification system: “Formulations of ingredients, mostly of exclusive industrial use, made by a series of industrial processes, many requiring sophisticated equipment and technology. Processes used to make ultra-processed foods include the fractioning of whole foods into substances, chemical modifications of these substances …”

That says nothing about whether it’s wrapped in plastic, which is a packaging rather than processing issue anyway. It also focuses on whether something is chemically modified rather than whether it’s in a typical kitchen. Many kitchens contain margarine, but that’s chemically modified; most won’t contain Balsam of Peru but that’s unmodified. Decaffeinated coffee is chemically modified, but to remove something (caffeine) rather than to insert an additive.

The definition concludes: “Processes and ingredients used to manufacture ultra-processed foods are designed to create highly profitable (low-cost ingredients, long shelf life, emphatic branding) convenient (ready-to-consume) hyperpalatable products liable to displace freshly prepared dishes and meals.”

Van Tulleken states that this part of the definition “suddenly resonated” with him and was the key to him embracing the NOVA classification: “to include, as part of the definition, the purpose for the processing – ‘to create highly profitable products’ – was completely new”.

But this is highly problematic as a definition. How can we know the purpose of processing, and thus classify something as a UPF – which is critical if we’re going to scientifically test the effects of UPF? In my own work on company purpose, I’m well aware of how researchers can define purpose to give whatever result they want. If the company performs well, claim it has a social purpose; if it performs badly, argue that it doesn’t. Similarly, if a food has negative health effects, declare it was produced to make money and therefore a UPF. Indeed, none of the implementations of the NOVA classification that I have been able to find aim to assess a profit motive. This seems to be something highlighted by van Tulleken to increase the book’s appeal to readers like me who are sceptical of food manufacturers.

And isn’t almost anything that a company does profitable? Companies are not charities; they’re for-profit organisations. This isn’t about greed but responding to customer demand; satisfying demand leads to profit. Companies make decaffeinated coffee in response to demand for a caffeine-free product; they make decaffeinated instant coffee to satisfy demand for convenient products. Other companies serve freshly chopped fruit wrapped up in plastic packaging, charging premium prices to customers who want pomegranate seeds for lunch on-the-go but don’t have the time to buy a pomegranate and chop it up. Beyond food, my gyms (Barry’s Bootcamp and F45) aim to provide top-flight – even addictive – classes, great customer service and a vibrant community, to retain existing clients and attract new ones. All of the above is motivated at least in part by profit, but this motivation is not at the expense of consumer health. What matters for health is the characteristics of the product, not the profit motivation – but van Tulleken stresses the latter as it feeds into the book’s anti-corporate appeal.

The profit criterion leads to one of the most bizarre UPF classification in the book. Chapter 16 discusses mineral water, which “my wife insists is the ultimate UPF” because a manufacturer “markets it aggressively for no other reason than financial gain”. Even though this is his wife’s classification, he doesn’t contradict it. This would suggest that #5 is sufficient for a food to be classified as UPF, in which case what it actually contains is irrelevant. But there’s no evidence that mineral water is bad for your health. Again, what matters is the product, not the producer’s motivation.

So just two pages into the definition, I am already confused as to what an UPF is. Is it:

Something wrapped in plastic packaging?
Something that contains at least one ingredient you wouldn’t find in your kitchen?
Something produced using sophisticated equipment and technology?
Something produced using chemical modification?
Something produced with the intent of being highly profitable?
Something convenient?
Something hyperpalatable?

#1 and #2 are the van Tulleken definition highlighted at the back of the book. #3-#7 are the NOVA definition, which is very different. Moreover, it’s not clear how to determine whether a food product satisfies the NOVA definition and thus counts as UPF. Even ignoring the ambiguity of the individual criteria (for #3, how do we know the intent behind a product and what counts as “highly profitable”?), does it need to satisfy all of #3-#7 or just some, and if so, how many?

And the confusion doesn’t stop there. Van Tulleken’s definition says that having a non-kitchen ingredient is the definition of UPF: if it “has one ingredient that you wouldn’t find in your kitchen, it’s UPF.” But elsewhere in the book he says that it’s simply an indicator of UPF: “the presence of those ingredients you don’t have in your kitchen is one indicator that a food is UPF”. In other words, it’s merely correlated with being UPF, just as shortness of breath is an indicator of pneumonia (but could indicate many other things). But if a non-kitchen ingredient is only an indicator of UPF, what is UPF?

a. Is It Processing?

The term “ultra-processed” suggests that processing, rather than ingredients, is the key feature of UPF, and this is what the book claims elsewhere. Chapter 11 highlights how “UPF is pre-chewed”, and this is what leads us to consuming too much because it’s so easy to eat. Every day after my workout I eat a smoothie containing whatever of kale, banana, oranges, apples, ginger, pineapple, grapes, and melon I have in the house, which I blend with a Nutribullet. It is processed so much that it’s no longer food (consistent with van Tulleken’s claim that UPF isn’t actually food); it’s a drink. By the “pre-chewed” definition, it’s UPF, but nothing has been added nor taken away. Being a drink means I consume more fruit than I would otherwise do, but this isn’t bad either. Oh, and I also put in a scoop of whey protein (one of the few UPFs I consume), which means that the whole smoothie should count at UPF. Yet protein aids recovery and is thus scientifically recommended post-workout, and it’s not clear that any harm outweighs the benefits of the kale and fruit.

b. Is It Additives?

Chapter 14 suggests that a UPF is something containing additives. This overlaps with #2 above but is far from the same. As discussed, there are non-kitchen items that don’t contain additives and kitchen items that do. And it’s not clear what an “additive” is – is it anything that’s added to a product? Van Tulleken suggests it’s “all those substances added to food for specific technological purposes”, but then a spice added to enhance flavour (thus making food more palatable and addictive) is an additive. But spices are found in a typical kitchen, so I’m confused. He also claims that spice extracts are UPFs, but it’s not clear why, since the production involves extraction not addition. Freshly-squeezed orange juice isn’t a UPF even though it’s extracted from oranges.

The book is equally confusing about whether additives are harmful or not. Chapter 5 states that “most additive anxiety is misplaced, as I’ll come to later”. Yet he comes to it, in Chapter 14 (“Additive Anxiety”), he describes in great detail the harm caused by specific additives, and stresses this is only the tip of the iceberg (“There are so many thousands of these additives that I won’t even be able to cover all the major categories. There are flavours, flavour enhancers, colours, emulsifiers, artificial sweeteners, thickeners, stabilisers, acidity regulators, preservatives, antioxidants …” and then list ten other categories). Yet Chapter 5 continues: “The main point is not that they are themselves harmful, its that the additives ae a proxy for UPF. They signal a method and purpose of food production that we now know is linked to disease”. So we are back to the nebulous and vague “purpose” classification.

c. Is It Sugar, Salt, and Fat Content?

As I’ll explain later, in other parts of the book van Tulleken highlights the sugar, salt, and fat content of UPFs. But the whole point of the book is that UPF is not about sugar, salt, and fat content but something different. Indeed, Chapter 7 is entitled “Why it isn’t about sugar”.

Yet the book is so contradictory that readers are confused also. The endorsement at the top of the Amazon listing is by James O’Brien: “A brilliant book . . . a real eye-opener. The bottom line is that if we keep consuming sugar, salt and fat on the scale that we currently do, it is going to cost all of us an unconscionable fortune in the future.” This contradicts the book’s entire thesis.

Ten Shades of UPF

So we now have three more definitions of what a UPF is for a total of ten. Is a UPF:

Something wrapped in plastic packaging?
Something that contains at least one ingredient you wouldn’t find in your kitchen?
Something produced using sophisticated equipment and technology?
Something produced using chemical modification?
Something produced with the intent of making high profits?
Something convenient?
Something hyperpalatable?
Something pre-chewed?
Something containing additives?
Something with a high level of sugar, salt, and fat?

The Book Views UPF as Black-and-White

Let’s put aside these inconsistencies and focus on van Tulleken’s definition. Criterion #2 is that it contains at least one ingredient that you wouldn’t find in your kitchen.

But such a criterion is extremely binary. It means that food is either UPF or non-UPF. A product with one non-kitchen ingredient is the same as one with 20. This is contradicted by van Tulleken himself in Chapter 5, when he says that “the individual ingredients of UPF may each be harmful, but it is in combination that they do the most harm”, which goes against such a black-and-white definition.

The definition also means that all non-kitchen ingredients are equally bad: spice extracts are the same as sodium nitrite (the preservative found in bacon, sausages, and ham). As the next section will discuss, there is no evidence for this.

The Evidence on UPFs is Weak

Such a simplistic definition of UPFs isn’t a problem if it’s backed up by evidence. Perhaps the evidence shows that containing even one non-kitchen item is harmful to health? But this is far from the case.

a. The Evidence Uses Different Definitions of UPFs from the Book

The first piece of evidence cited on the harm from UPFs is a paper by Kevin Hall et al. in Cell Metabolism. That it appears first is deliberate; van Tulleken considers this a key piece of evidence and refers to Hall many times throughout the book.

The study takes just 20 patients, and gives them diets with the same calories, sugar, fat, fibre, and macronutrients but differing according to whether they contain UPF, defined by the NOVA system. Let’s ignore the classification issues identified earlier (whether a food product needs to satisfy all of #3-#7 or just some) and the tiny sample size. This study is only evidence on UPFs classified by NOVA. It has no bearing on UPFs classified by van Tulleken – which is the definition highlighted on the back cover and in the introduction, and is what the book is famous for.

Now this wouldn’t be a problem if van Tulleken’s definition were highly correlated with NOVA’s – if #1-#2 were a convenient proxy for #3-#7. But van Tulleken offers no evidence for this. His definition simply appears out of thin air. I looked into several of the other studies that he cites and couldn’t find any that classified UPFs by plastic packaging or inclusion of one non-kitchen item. This is the problem highlighted in Chapter 3 of May Contain Lies (“A Statement is Not Fact”) – studies being quoted for something quite different from what they actually study.

b. The Book Overextrapolates From Narrow Evidence

Of course, the Cell Metabolism study is far from the only evidence cited in the book. In Chapter 14, it cites a paper in Nature on how carboxymethylcellulose and polysorbate 80 cause harm. This study analyses mice, and elsewhere in the book van Tulleken criticises studies contradicting his conclusions for being on mice rather than humans. Let’s ignore this problem. Van Tulleken declares that “The conclusion to the Nature paper proposed that dietary emulsifiers ‘may have contributed to the post-mid-twentieth-century increase in incidence of inflammatory bowel disease, metabolic syndrome, and perhaps other chronic inflammatory diseases’.”

But van Tulleken selectively quotes only half a sentence. The full quote is the following: “While additional studies will be needed to determine if CMC, P80, and/or other emulsifiers impact human health, our observations in mice suggest the possibility that dietary emulsifiers may have contributed to the post-mid 20th century increased incidence of IBD, metabolic syndrome, and perhaps other chronic inflammatory diseases.” In addition to “while additional studies will be needed … suggest the possibility” being important caveats, the omitted first half stresses how it only studies carboxymethylcellulose and polysorbate 80. It has no implications for other emulsifiers. Moreover, a study on emulsifiers has no implications for other additives, or other non-kitchen items in general. This is an example of the “granularity” issue in Chapter 2 of May Contain Lies: we overextrapolate from a tree to draw conclusions about the whole forest, when the world is granular.

The next sentence in the Nature paper is also interesting: “Importantly, this hypothesis does not dispute the commonly held assumption that excess caloric consumption is a predominant factor driving the metabolic syndrome epidemic”. It goes against the book’s entire thesis, so van Tulleken conveniently ignores it.

c. The Book Mixes Up Correlation and Causation

The author claims to be well aware of the differences between correlation and causation – does harm result from food itself, or other factors associated with food consumption? However, much of the evidence he quotes fails to disentangle the two.

After (correctly) recognising that the Hall study was not definitive, as it’s a single study conducted in a lab rather than real-world setting, van Tulleken resolves to do his own research: to eat a 80% UPF diet and see what happens. But there are many problems with this. Most obviously, it’s a single datapoint; as highlighted in Chapter 4 of May Contain Lies, “a fact is not data: it may not be representative”. Perhaps many other people have started UPF diets (out of convenience or affordability) and did not suffer negative consequences, and so had no books to write.

More importantly, it’s impossible to attribute any consequences to the UPF diet alone. As the author acknowledges, in the new regime “I’d just eat when I felt like it”. Thus, the volume of fat, sugar, and salt will have significantly increased, contrasting his claim to isolate the effect of UPFs and hold these other factors constant. Moreover, “I’d just eat when I felt like it” may have caused him to embark on a generally sloppy lifestyle. Perhaps he also lay on the couch whenever he felt like it (rather than exercising), went to bed whenever he felt like it, and drank alcohol whenever he felt like it. Indeed, he acknowledges that he started to engage in sloppy parenting, allowing his daughter to eat his junk food also, so the new regime led to general sloppiness.

An MRI scan at the end of his new regime showed that his brain was firing in different ways; in particular, his self-control function was weaker. But that could be just due to general sloppiness, rather than UPF in particular. If the new regime involved playing video games, watching TV, or scrolling through social media whenever he felt like it, his self-control might have similarly been weakened.

Another case study of one that van Tulleken cites is his brother moving to America and living in a “food desert” where UPFs are widely available. But, as van Tulleken acknowledges elsewhere, food deserts are highly correlated with poverty, and that could be at least partially responsible for his brother’s obesity. He also highlights that his brother “was also stressed, far from home, and undergoing huge changes in his life. Stress, from any source, but especially the chronic stress of poverty, has dramatic effects on the hormones that regulate appetite, increasing the drive to eat.” Yet he tosses aside the alternative suspects of poverty and stress, and attributes everything to his preferred scapegoat of UPFs.

d. Many consequences attributed to UPF may have nothing to do with UPF

Even if you can pinpoint the food itself (rather than poverty, stress, or sloppiness) as being the cause, it’s not clear which aspect of the food is the culprit. Let’s go back to the Kevin Hall et al. study in Cell Metabolism. It takes 20 people, gives half of them one diet, and gives the other half an identical diet in terms of fat, salt, sugar, and fibre but with UPF. Crucially, “participants could each as much as they liked”, so any outcomes likely come from greater overall consumption – and thus greater intake of fat, salt, and sugar – rather than the UPFs themselves.

But if the UPFs lead to the greater consumption, then they can indeed be blamed. However, this is far from clear. As van Tulleken highlights throughout the book, UPFs make food tastier – indeed, hyperpalatable according to criterion #7. The study simply shows that, if food is tastier, people eat more of it. Adding natural spice to food makes it tastier without altering the fat, salt, sugar, and fibre content, and may well have also increased consumption, but it’s not UPF.

Chapter 7 highlights how sugar leads to tooth decay and tirelessly goes through all the reasons why tooth decay are bad. But this is not because sugar is a UPF due to the refining process; it’s because sugar contains, well, sugar. Honey is not UPF but also causes tooth decay due to its sugar content. In fact, it can be even worse because it sticks to your teeth, maximising the time your teeth are in contact with it. Highlighting all the problems with sugar contradicts the entire premise of the chapter, which is entitled “Why it isn’t about sugar.” But some things are.

Similarly, Chapter 2 decries the high salt content of Coco Pops: “The incredible saltiness is true of most breakfast cereals – it helps to make them taste amazing. So, why isn’t there a salt warning?” But the problem is salt, not UPFs. According to his definition, salt is not a UPF as it’s found in every kitchen. So is the problem salt or UPFs? Indeed, UPFs could actually reduce salt consumption by being an alternative way to enhance taste. And food does have salt warnings: the red, amber, and green traffic lights explicitly concern salt, yet van Tulleken decries them since he argues that UPFs matter, not salt, sugar, or fat content.

Chapter 17 tackles the environmental impact of UPFs. Beefburgers certainly have a significant carbon footprint, but this isn’t because they are processed – it’s because they contain beef. A raw beef steak, which is clearly not UPF, also has a significant carbon footprint. Two chapters later, in yet another attempt to define UPF, he writes “The NOVA classification has forced me to consider the purpose of the food the whole time. … Is it sitting within a food system that causes climate change.” What does “sitting in a food system” even mean? Doesn’t all food sit in our food system? Any meat or dairy product has a significant carbon impact, but much of this is not UPF.

Back to Chapter 17, Van Tulleken talks about the issues with battery farming – not just environmental, but also welfare – but again this has nothing to do with whether the chicken is processed afterwards. Focusing on the additives themselves, it’s true that some have an environmental impact but not others (such as pectin, derived from fruit peels – a by-product – and biodegradable). Some may have a net positive impact, if preservatives extend the shelf life of food and lead to less being thrown away, but the author ignores this.

The same chapter correctly highlights the danger of anti-microbial resistance. But again this is not UPF. He conflates two separate issues – how animals are reared and what their meat is used for afterwards. It’s as if UPFs are his hammer and everything is a nail; he wishes to attribute everything that’s wrong with the food industry to UPFs.

The Evidence on Other Factors is Weak

The book engages in black-and-white thinking along several dimensions. It is black-and-white in claiming that all non-kitchen items are bad, when the evidence is specific to a few. It is also black-and-white in claiming that UPFs (plus genetics) are the only driver of obesity: willpower and exercise have no effect. This leads to the compelling message that obesity isn’t your fault. But the evidence cited is far weaker than the author claims.

a. Exercise

Van Tulleken selectively cites studies suggesting that exercise has no effect on obesity. But just as he failed to hold other factors constant when pinpointing the effects of UPFs, he also fails to do so when identifying the effect of exercise. These studies consider people who engage in physical activity and, according to van Tulleken, think “because I’ve worked out, I’ll be lazy the rest of the day”. This is an invalid comparison. Since exercise burns more calories than being at rest, all else equal it automatically attenuates obesity.

Not holding all else equal leads to the absurd claim that “people burn the same amount of energy each day whether they walk ten miles or sit at the desk” which is scientifically incorrect, that “we cannot lose weight just by increasing activity” (emphasis in the original), and that coal miners and athletes burn the same number of calories as couch potatoes (in contrast to scientific advice on recommended calorie intake). Others have highlighted other basic errors in van Tulleken’s quoting of these studies; for example, when claiming that miners burned 2,100-2,800 calories per day – the same as everyone else – he only includes the calories burned at work. Their actual daily burn was 3,658.

Moreover, all else is unlikely to be equal, as exercising causes a shift in mindset and leads to better self-control in many other areas, including food consumption: see the evidence cited in Spark: How Exercise Will Improve the Performance of Your Brain by John Ratey and Eric Hagerman and my Gresham College public lecture on Mental and Physical Wellness.

And black-and-white thinking rears its ugly again when van Tulleken uses very narrow studies, such as on coal miners, to rubbish all forms of exercise. Some types, such as Low Intensity Steady State, are designed to overcome the problem of being too tired after exercise that you do nothing for the rest of the day. Weight training increases your resting energy expenditure since muscle tissue is more metabolically active than fat.

b. Willpower

There is widespread evidence that willpower matters over and above genetics, such as the work of Carol Dweck on mindset or Angela Duckworth on grit (in certain contexts). Rather than citing scientific research, this claim is largely based on “I know this from personal experience”. He claims that his brother could do nothing to overcome the food desert that was Boston. He argues that there was nothing he could do to stop his kids wanting to eat his food when he was on his UPF diet: “It was impossible to stop the kids from eating my Coco Pops, slices of pizza, oven chips, lasagne, chocolate. If I went to eat secretly, Lyra would hunt me down and demand whatever I was having”.

But you can’t draw sweeping conclusions from one person’s experience. Just because van Tulleken lacked the willpower, it doesn’t mean that others do too. I lived in Boston for four years on a student budget during my PhD. It’s perfectly possible to stop your kids eating junk food: you just don’t give it to them. If your kids demand to drink the wine you’re drinking, you wouldn’t oblige, and it’s the same with UPF. Van Tulleken could have eaten the chocolate at night when Lyra was asleep. Poor parenting is not an excuse to blame companies.

It’s also not hard to instil willpower in kids. My son has just turned 3 and knows that he’s only allowed one biscuit a day. If I offer him a biscuit in the afternoon, he’ll turn it down if my wife gave him a biscuit in the morning (which I didn’t know about), saying that more than one treat a day will give him black teeth. But van Tulleken did not attempt to instill any such willpower in Lyra. He gave her as much Coco Pops as she wanted: “Lyra put her ear to the bowl and shut her eyes, entranced. She then began to eat again. And eat. And eat. As I watched her, it seemed she wasn’t fully in control. … Her eating wasn’t just mindless: it was trancelike”. To avoid mindless behavior, we exercise our mind.

Such is van Tulleken’s desire to abdicate personal responsibility that he shuts off any critical thinking faculties and makes unsupported logical leaps. Referring to Coco Pops he writes “there was that cartoon monkey on the box, suggesting that the cereal was not merely safe for kids, but deliberately intended for them. Maybe it was fine.” It’s not clear why he uses the indicator of a cartoon monkey as a signal that kids should consume the product without limit (the chapter is entitled “I’d rather have five bowls of Coco Pops”). If an advert for whisky contains a white man, does this suggest that its safe for all white men to drink as much of that whisky as they want? Later he argues “an addictive substance being marketed with a monkey to my three-year-old is a regulatory failure”. No, purchasing it and allow your three-year-old to each as much as she wants is parental failure. Three-year-olds don’t make purchasing and consumption decisions; parents do.

Beyond van Tulleken’s or my personal experience, the claim that you can do nothing about UPFs is first-order rejected by the facts. People who live in the neighbourhood and thus have the same access to UPFs end up eating very different diets. The same is true for siblings who share the same genes and live in the same house, or people attending the same buffet. The environment certainly matters. But to claim that it’s the only thing that matters, and that willpower can’t help you because companies’ additives are a dark art and black magic that overwhelm it, is incorrect.

The Book Offers No Practical Takeaways

After the reader through over 300 pages on the dangers of UPF, the final chapter, “What to do if you want to stop eating UPF” is embarrassingly short, at just two pages. Indeed, it’s difficult for it to convincingly recommend “stop eating UPF” since the entire premise of the book is that you can’t stop – willpower doesn’t work because you’re the helpless victim of evil corporations. Indeed, it seems there is no point reading the book except to make you feel better that obesity is not your fault. The back cover claims that it is a “manifesto to change how you eat”, but the book declares that you cannot change. One might suggest that the logical takeaway is to eat less UPF, but he earlier rubbished this suggestion: “simple advice like ‘Eat less and move more!’ is ineffective for sustained weight loss. It’s as crazy as saying ‘Drink less water to someone who’s feeling thirsty’ ” (itself a crazy comparison as water is good for someone who’s thirsty, but more food is bad for someone who’s obese).

Much more useful would have been to highlight the specific additives that are harmful, and the ones that are not. But that wouldn’t have led to the simple black-and-white rule of “avoid everything wrapped in plastic and containing one non-kitchen item”, the widespread indictment of food companies that appeals to many readers, or to mass alarm and panic which leads to endorsers claiming that everyone needs to go out and buy this book. Indeed, to create even more panic, van Tulleken highlights a consequence that might be even more important than obesity to some readers “the plastics from UPF packaging … may even cause penile shrinkage” – but doesn’t cite any evidence.

It would have also been helpful to explain how much of an additive you can consume without harm given that, as the author claims, UPFs are very difficult to avoid. Just saying “avoid UPFs” is as useless as “avoid air pollution”. If you live in a city, you’ll be exposed to air pollution – which is why air quality monitors advise you when pollution is so high that you should stay indoors. Some people could respond to the book by hiring a housekeeper and/or private chef to enjoy home-cooked meals every day. But others aren’t so fortunate, and so it’s difficult to do anything with blanket advice to “avoid UPF”.

The penultimate chapter is much longer, and argues for more regulation. While this is less practical because readers don’t directly set regulation, they can indirectly affect it by encouraging or voting for politicians. I’m a supporter of correctly designed regulation (see Chapter 10 of Grow the Pie) to correct market failure, and specific UPFs that create negative externalities are indeed a market failure.

But this chapter again lacks any nuance and presents a black-and-white view of regulation as having only benefits with no costs. He argues that the burden of proof should be on companies to show that UPFs cause no harm (“absence of evidence is not evidence of absence”) rather than activists to show that they do cause harm. But this is inconsistent basic regulatory principles. Inventors of calculators did not need to show that their introduction doesn’t harm numeracy by reducing the learn to use your times tables, nor did Google prove that its search engine doesn’t discourage the need to learn facts. In the specific domain of food, Nobel Laureate Sir Richard Roberts and others have repeatedly argued that Europe’s over-regulation of genetically modified food has led to the deaths of millions by denying them the technology that would have fed them. As the saying goes, “The US innovates, China replicates, and Europe regulates”, and such an approach to regulation would stifle many innovations that improve welfare.

Rather than blaming companies or leaving it to regulators, the book should empower the reader by providing actionable advice. But it does not. It disenfranchises the reader by abdicating personal responsibility.

How Did We Get Here?

So how did we get to a book which makes such sweeping, exaggerated conclusions, much more extensive than those warranted by the data? Throughout his book, Van Tulleken liberally points out the potential conflicts of interest from the funding of scientific research, but never acknowledges his own – quite substantial – conflict of interest.

Where does his funding come from? His readers. He wants to write what sells, and playing into confirmation bias (corporations are evil, obesity is not your fault) and black-and-white thinking (all UPFs are bad, all the harm from the food industry is caused by UPFs, nothing can save you from them) does exactly that. Such a conflict gives a significant incentive to overexaggerate and overextrapolate his claims, and ignore the shades of grey in complex issues.

Now that doesn’t mean that every book is misleading. Some authors are aware of the incentives but won’t respond to them; they’ll stay true to the evidence. But equally, this does not mean every scientist who is reimbursed expenses by a food company will abandon their professional principles. What matters is the content of the book or study, which is what my commentary has focused on, whereas van Tulleken argues that many scientists are automatically guilty by association.

Conclusion

Overall, the book is valuable in highlighting that certain additives are harmful and that your health depends on your consumption of these additives, not just sugar, fat, and exercise. But the claim that all non-kitchen ingredients are harmful, the over-attribution to UPFs of issues that may have a myriad other causes, and the charge that nothing else – will power, exercise, or reduced consumption of other nutrients – can save you, is not supported by the evidence. The most dangerous aspect of the book is the abdication of personal responsibility and the desire to blame others for decisions – consumption, exercise, parenting – that we take ourselves.

An Addendum on Infant Formula

The discussion on breast- vs. bottle-feeding (in Chapter 6 of May Contain Lies) is the topic about which I’ve received most reader/listener feedback on, interestingly given I’m a finance professor. Many said that they felt substantial pressure to breast-feed and were stigmatised when they chose not to, based on apparent cast-iron evidence that bottle-feeding will harm their kids’ physical and mental development. One (who gave me permission to share) is expecting her first baby and previously had a double mastectomy due to breast cancer. Her hospital, a leading medical provider on the East Coast of the US, repeatedly stressed the criticality of breastfeeding because of evidence that it benefits both the mother and baby. Indeed, it believes the evidence is so strong that it does not provide resources for mothers who are unable to, or choose not to, breastfeed – it focuses on recommending breastfeeding only. The listener shared that she was brought to tears when hearing the actual evidence in my book, in contrast to what was being dictated to her.

I thus found the coverage of infant formula in van Tulleken’s book to be both unevidenced and unhelpful. He claims that “in the UK the influence of industry over every aspect of infant feeding means that the women who want to breastfeed, there are barriers and a lack of support”, and that “the current environment in the UK is stigmatizing for all mothers and the marketing harms everyone.” If anything, it’s the other way: there is a lack of support for bottle- or combination feeding. The National Childbirth Trust steadfastly recommends exclusive breastfeeding. When I called their helpline, after my son had exhausted my wife’s breastmilk and was still hungry, their representative was adamant that I should not give him any formula, before admitting at the end that “I’m not medically qualified”.

He claims that “limiting the marketing of formula would be the single most effective intervention for the prevention of deaths in children under five years of age¹⁷”. The superscript 17 suggests that it’s backed up by evidence and so it must be true. But footnote 17 doesn’t say anything of the sort. It’s a letter, not a peer-reviewed study, and is about the proportion of UK women who are breastfeeding a year after birth – not whether formula increases deaths, and certainly not whether banning the marketing of formula would reduce more deaths than, say, discouraging parents from smoking or putting newborns to sleep on their front. For me and many of my readers (and for the far more numerous readers of Emily Oster’s work), learning that formula can be used in combination with breastmilk was very valuable.