Michael: In 1998, a scientist named Roy Baumeister decided to investigate a question that psychologists had been thinking about for a while: Do people have a limited amount of willpower?

To figure it out, he and his research group did what researchers do: they conducted a study. The results of that study seemed to show that humans do have a limited pool of self-control. In other words, once we’ve had to resist temptation, it’s a lot harder to do it again. They called this effect ego depletion, and the idea has had a huge influence on psychological research ever since. It’s been incorporated into things like dieting tactics, athletic training techniques, and advertising.

But now, it seems like the phenomenon might not exist at all -- which has a lot of scientists worried about the way we do psychology research in general.

[SciShow intro plays]

The concept of the ego goes way back. Sigmund Freud -- who’s often called the father of modern psychology -- described the ego as the aspect of our personality that makes decisions about how to interact with the real world. The ego has to figure out how to balance our needs and desires within the context of reality and society. Freud believed that the ego would need energy to keep up with those demands and decisions, but that was about as far as he got.

Baumeister and his colleagues were trying to test the 1990’s-version of this idea: that self-control is a limited resource, and it takes energy and motivation to maintain restraint. According to this ego depletion theory, every time you use your self-control, you draw on that strength, and it takes some time for you to recover it. So, to test this hypothesis, Baumeister’s team decided to create a task that would require two acts of self-control back-to-back. They figured if self-control can be used up, then having to use a lot of self-control on the first task would mean that the subjects would have less self-control available for the second task.

So first, they baked a bunch of chocolate chip cookies in a little oven in the lab, so the whole room smelled like fresh cookies. Then, they invited 67 psych students into the lab, making sure they hadn’t eaten anything for a while. When the students arrived, they were each brought into the baking room and asked to sit at a table. On the table were two plates: one filled with those fresh, delicious cookies, and the other filled with radishes. Some of the students were told that they had to eat the radishes -- but no cookies -- while others were told that they were supposed to eat the cookies -- but no radishes. Then, each student was left alone in the room for five minutes. When the experimenter came back, they asked the student to try to solve a difficult puzzle.

But here’s the thing: the puzzle was designed to be impossible to solve. What the researchers really wanted to know was how long it took for the students to give up -- and if spending a few minutes resisting cookies would make it harder for them to keep trying. They found a big difference between the students who weren’t allowed to eat cookies and the students who were. The students who were asked to only eat radishes before doing the puzzle gave up on it in about eight minutes. But the students who were told to eat the cookies worked at the puzzle for more than twice as long -- almost nineteen minutes on average. When a third group of students did the puzzle test without any cookie encounters at all, they spent only a little longer trying to solve it, averaging 21 minutes each.

This seemed like very solid evidence for ego depletion. The researchers concluded that the students who were only allowed to eat radishes had to use a lot of self-control to not touch the delicious, chocolatey cookies, so when they got to the puzzle test, they didn’t have much willpower left to spare. But the students who got to eat the cookies didn’t have to control themselves -- they could just dig right in. Their willpower reserves were much higher, so they worked at the puzzle longer.

The same group did several more experiments designed to test willpower under different circumstances and found the same results -- whenever participants were asked to do something that required self-control, they had a harder time with the puzzle task. Which was a huge deal. A lot of the time, it can be difficult to draw broad conclusions from psychology studies. Often, effects are only seen under very specific conditions, and if you change an experiment even a little, you might end up with very different findings.

But Baumeister and his colleagues had shown that this effect applied to more than just cookies and radishes -- ego depletion showed up in all kinds of different experiments. After that, ego depletion basically became its own subfield of psychology. In 2007, researchers at Florida State University figured out what seemed to be happening, biologically, as people used up their self-control: their blood sugar levels were dropping.

Our brains are high-energy organs: they need a lot of energy to keep going. And that energy comes from sugar. The team thought that the psychological effort of maintaining self-control might use up extra energy, which could mean that the brain was burning through its available reserves faster than normal. They tested this idea by having some subjects do willpower-related tasks, like watching an emotional or graphic video without showing any emotion, while other subjects didn’t have to hold back. Sure enough, the people who’d used some willpower had lower blood glucose levels. And! When they replenished that glucose with a nice, refreshing glass of sugary lemonade, it seemed to restore their self-control.

The evidence was pretty convincing -- and became even more convincing in 2010, when yet another group of researchers did a meta-analysis. They examined the results from 83 different published studies on ego depletion, and concluded that the effect was real. But recently, some studies have started to cast doubt on this whole ego depletion thing.

One group of researchers, for example, found that subjects didn’t have to actually drink lemonade to replenish their willpower -- just tasting it, without the corresponding changes in blood glucose, could be enough. Others found that what subjects believed about willpower could affect their performance on self-control tasks. Participants who believed ego depletion was real gave up on difficult tasks way before those who didn’t. And, people who thought of willpower as infinite resource did just as well on the second difficult task as people who hadn’t done the first task at all.

Then, in 2014, some scientists tried to replicate the original studies and couldn’t find the effect. The group decided to look more closely at that 2010 meta-analysis, and they found a whole bunch of problems. For one thing, when they reanalyzed the data using newer methods, the ego depletion effect disappeared. For another, the data included in the analysis in the first place was probably biased in favor of ego depletion. See, scientific publications tend to publish positive results -- the kind that prove the researchers’ hypothesis. In other words: researchers might have done lots of studies that didn’t find evidence of ego depletion -- but odds are, those results wouldn’t have gotten published.

The meta analysis only looked at the data from published studies -- so there might have been plenty of studies that didn’t find the ego depletion effect, but the analysis just wasn’t taking those into account. When the team went back and included other, unpublished data in the analysis, they found that -- again -- the ego depletion effect just wasn’t there. These results were popping up right as a wave of concern about replicability was spreading through the psychology community. More and more researchers were saying that they couldn’t reproduce the results of various psychological studies.

This wasn’t just ego depletion, this was all of psychology, and people were starting to doubt whether they could really trust the published literature. All of this uncertainty led the Association for Psychological Science to open a Registered Replication Report on ego depletion. Basically, one official experiment -- based on a study originally published in 2014 -- would be conducted by researchers in many different labs. Each group of scientists would analyze their results and compare them to the results of the other research groups.

Baumeister, the head of the original study, was brought in to help design the experiments, and the psychologist who led the 2010 meta-analysis, Martin Hagger, agreed to lead the project. Subjects were told to watch words flash on a screen and to press a button whenever they saw a word containing the letter e - but not if the letter e was within two spaces of another vowel. If that sounds like the most tedious, annoying thing ever -- that was the point.

The subjects had to pay close attention, and just like the original study, they then had to perform a follow-up self-control task. But when Hagger’s team did the same experiment in 24 different labs in different countries and languages, the ego depletion effect evaporated. Only two of the 24 groups found the effect at all -- and even one group found the opposite effect!

So what does all of this mean for ego depletion? It depends. You could argue that it’s just this particular task -- with the words flashing on the screen -- where ego depletion doesn’t show up, but that willpower is still a limited resource. The thing is, the main reason why the ego depletion effect was such a big deal was that it held true across so many different tests -- from cookies and radishes to conversations about race and discrimination to making decisions about what cleaning product to buy. But now, it seems like ego depletion only happens under very specific circumstances -- if at all.

Baumeister has said that he plans to launch his own replication studies, in the hopes of proving that the effect is real. And other scientists who participated in the Registered Replication Report have said that this single failed replication isn’t enough -- more research needs to be done with different experimental methods before psychologists can come to any real conclusions. In the meantime, some researchers have described the field as being back at square one: ego depletion might be real, but it’s going to take a lot more work -- and a lot more replication -- to know for sure.

Either way, this whole thing is serving as a kind of cautionary tale when it comes to scientific research, especially in social psychology. Often, sample sizes are small, and the subjects are mainly university students -- so results of studies don’t necessarily apply to the whole population. It’s also very difficult -- if not impossible -- to reproduce the results of earlier research in other labs, with different researchers and subjects. For scientists, who are just looking for the truth, it’s a problem. And lots of people are talking about ways to improve the situation.

One thing that might help is changing the way experimental results are analyzed and published. This could involve publishing more negative results, when the outcome of a study doesn’t prove the researchers’ hypothesis. It could also include more open sharing of data and analytical methods, so that other, independent researchers can verify results. Hopefully, with changes like these, we’ll end up with more reliable results and better science.

Thank you for watching this episode of SciShow, if you want to help us keep bringing you episodes like this, go to patreon.com/scishow. And don’t forget to go to youtube.com/scishow and subscribe!