Why you keep eating what’s in front of you even after you’re full

This is Science Fictions with Stuart Ritchie, a subscriber-only newsletter from i. If you’d like to get this direct to your inbox, every single week, you can sign up here.

Imagine you’re eating a bowl of tomato soup. Depending on how hungry you are, you’re either going to get to the end of the bowl or stop somewhere along the way because you feel full.

“But what if,” say the authors of a new study, “the laws of physics had glitched, and this bowl had been refilling itself with soup after every spoonful you took?” Yes, it’s a strange thought, but go with it: the question is whether this would cause you to eat more soup than usual. Would you be relying on your internal feeling of how full you are or on what you see in front of you?

This is a more interesting question than it first seems: if we want to help people lose weight, or just eat more healthily in general, we need to know as much as we can about the psychology of when we feel satiated.

And that’s where the self-refilling soup bowl comes in. The scientists behind the new study really went to the trouble of making one: they rigged up a series of pipes under a table that were secretly connected to a soup bowl. As soup was taken out of the bowl, it automatically filled up with more from a big pot at the end of the table.

It’s a wonderfully creative – if bizarre – set-up for an experiment on food psychology. But, before we get to the results, we need to take a little detour into the career of the man who originally came up with the bottomless soup bowl study: Brian Wansink.

The rise and fall of Brian Wansink 

Wansink was a psychology professor at the prestigious Cornell University in the US. He and his “Food and Brand Lab” produced lots of well-cited scientific publications, and he advised the US government on its food strategy. He could regularly be seen in YouTube videos talking about his weird and wonderful studies, like the one where his team measured the angle of the eye-gaze of cartoon characters on cereal boxes and found they were looking downwards, presumably towards children to catch their interest. 

You’ve probably heard about some of his studies before: if you are aware of the belief that people will eat more if you give them large plates at a buffet, then you know about Wansink.

He published the original “self-refilling soup bowl” experiment in 2005. The experiment won him an “Ig Nobel Prize” – the tongue-in-cheek scientific award for studies that “make us laugh, then make us think”.

So Wansink was great at getting his ideas out there. The problem was that he wasn’t very good at the actual science. In 2016, he wrote a blog post that explained how he taught his students to do research. Essentially, he encouraged them to dredge through a dataset, “trying” lots of different hypotheses, then publishing only those that showed the results he wanted.

This, of course, is not what we want scientists to be doing: if their hypothesis doesn’t work out, they need to be honest about that and publish the “null” finding. After all, the fact a hypothesis wasn’t supported by the data is still an advance in our knowledge of the world. And hiding results does the scientific literature a major disservice.

Worse, though, it means that the results you do publish might be false positives: just spurious statistical flukes that nobody will be able to replicate. In this sense, Wansink’s ill-thought-through blog post was seen as emblematic of the “replication crisis” in science.

Lots of scrutiny followed. Scientists looked into Wansink’s research papers and found a huge amount of highly sloppy work. In total, he ended up having to retract 18 of his studies from the literature, and even more of them had to be corrected.

Cornell investigated and found that he had committed many kinds of scientific misconduct, including “misreporting of research data, problematic statistical techniques, failure to properly document and preserve research results and inappropriate authorship”. He resigned from his professorship soon afterwards.

All of which leaves us with an interesting question. Were any of his results real? After all, just because he committed misconduct when running and reporting the experiments, it doesn’t necessarily mean that someone else couldn’t find the same results if they ran his studies properly.

And that’s just what’s happened with the bottomless soup bowl study.

A replication 

The new study had a much bigger sample size than Wansink’s original: 464 people compared to 54. But, otherwise, they tried to make it as similar to his study as they could – the exact same refilling-bowl set-up, and the same questions asked to the participants about how much they ate and how hungry they felt. They were also able to measure, from the soup pots, exactly how much each person had eaten.

Here’s what they found: the people who ate from the self-refilling bowls consumed 32 per cent more soup than those who ate from normal bowls (which were manually and obviously refilled by the researchers such that the participants could clearly see they would be eating more). This was a statistically significant effect and went in the same direction as the result from the original Wansink study.

In Wansink’s study, the self-refilling group had eaten 73 per cent more soup – a much bigger effect. It is not clear why that would be, misconduct of some form in the original study? Just random statistical fluctuation? But the fact remains that the basic finding did replicate. After all that, Wansink was right.

The crucial thing is that, in both experiments, the people eating from the self-refilling bowl ate that extra soup without noticing that they’d done so (indeed, in the new experiment they excluded data from 190 people who worked out what was happening and so spoiled the set-up of the study). It’s evidence that, at least to some extent, people eat mindlessly: they’re not really aware of how many calories they’re consuming just from internal feelings of fullness or satiety, and they can be confused by outside indications such as how much soup is in their bowl.

That’s interesting and potentially important for people planning diets or health interventions. But it’s also a good lesson for those of us who are interested in the dodgy side of science: just because a researcher ends up in disgrace for bad research doesn’t mean that all their studies should necessarily be trashed.

I was amazed to see that Wansink’s results, which I had assumed were all complete rubbish, had been replicated in this entirely independent investigation. I’m still a little sceptical overall (for instance, did more than the 190 people notice, or at least suspect, what was going on but just not say anything?), but I’ve had to update my mental model towards this much-ridiculed study actually being a worthwhile addition to the scientific literature.

But wouldn’t it be nice if we could have trusted it to begin with? If Wansink himself had run it again to check whether it wasn’t a fluke? If anyone else in the past nearly 20 years had thought to replicate it, without having the extra impetus of Wansink’s resignation pushing them towards doing so?

Bottomless soup bowls are real, but a truly rigorous science where researchers take replication seriously is still a way off yet.

Science link of the week

Actually, the secret is that I’m not here this week – I should be in Australia as you’re reading this. So, there could be all sorts of developments in science this week but I don’t know what they are. The newsletter will take a week off next week and will return on 7 December.

See you then!

This is Science Fictions with Stuart Ritchie, a subscriber-only newsletter from i. If you’d like to get this direct to your inbox, every single week, you can sign up here.