In 1998, the psychologist Roy Baumeister introduced the “strength” model of self-control. It’s a slightly misleading name, since the model attempts to describe the weakness of the will, why people so easily succumb to temptation and impulse. In Baumeister’s influential paper – it’s since been cited more than 2500 times – he and colleagues describe several simple experiments that expose our mental frailties. In one trial, subjects forced to eat radishes instead of chocolate candies gave up far sooner when asked to solve an impossible puzzle. In another trial, people told to suppress their emotions while watching a tragic scene from Terms of Endearment solved significantly fewer anagrams than those who watched a funny video instead. The lesson, writes Baumeister et al., is that the ego is easily depleted, a limited resource quickly exhausted by minor acts of self-control.

It’s a compelling theory, as it seems to explain many of our human imperfections. It’s why a long day at work often leads to a pint of ice cream on the couch and why we get grumpy and distracted whenever we miss a meal. Because the will is so feeble, we must learn to pick our battles, exerting power only when it counts. In a particularly clever follow-up experiment, published in 2007, Baumeister and colleagues showed that a variety of typical self-control tasks led to lower glucose levels in the blood. (The mind, it seems, consumes more energy when attempting to restrain itself.) Not surprisingly, giving “depleted” subjects a glass of sweet lemonade improved their subsequent performance on yet another self-control task. However, depleted subjects given lemonade sweetened with fake sugar experienced no benefits. Saccharine might trick the tongue, but it can't help your frontal lobes.

So far, so depressing: as described by Baumeister, self-control is a Sisyphean struggle, since the very act of exerting control makes it harder to control ourselves in the near future. We can diet in the morning, but that only makes us more likely to gorge in the afternoon. The id always wins.

But what if the will isn’t so fragile? In recent years, several papers have complicated and critiqued the strength (aka ego depletion) model of self-control. In a 2012 paper, Miller et al. pointed out that only people who believed in the impotence of willpower – they agreed that “after a strenuous activity, your energy is depleted and you must rest to get it refueled again” – performed worse on repeated tests of the will. In contrast, subjects who believed that self-control was seemingly inexhaustible – “After a strenuous mental activity, you feel energized for further challenging activities” – showed no depletion effects at all. This suggests that the exhaustion of willpower is caused by a belief about our mental resources, and not by an actual shortage of resources. We think we’re weak, and so we are. The science becomes a self-fulfilling prophecy.

That’s a long introduction to the latest volley in the ego-depletion debate. In a new paper published in the Journal of Personality and Social Psychology, Veronika Job, Gregory Walton, Katharina Bernecker and Carol Dweck left the lab and tracked more than 100 students at a selective American university. The assessment began with a survey about their willpower beliefs. Is it depleted by strenuous mental activity? Or does “mental stamina fuel itself”? Then, the students were sent weekly questionnaires about their self-control failures in a variety of domains, from academics (“How often did you watch TV instead of studying?”) to emotional control (“How often did you have trouble controlling your temper?”) Finally, Job, et al., asked students to anticipate the amount of self-control they’d need to exert over the next week. Did they have a big exam coming up? A class presentation? Were they having problems with friends or professors? In addition to these surveys, the scientists got access to the students’ GPA.

When the demand for self-control was low, the students’ beliefs about willpower had no effect on their self-control performance. However, when the semester got stressful, and the students felt a greater need to resist temptation, the scientists observed a significant difference: those who believed they had more self-control were better able to control their selves. Here are the scientists: “Far from conserving their resources and showing strong self-regulation when needed, students who endorsed the limited theory [of self-control] and who dealt with high demands over the term, procrastinated more (e.g., watching TV instead of studying), ate more junk food, and reported more excessive spending as compared to students with a nonlimited theory about willpower.” (This relationship held even after controlling for trait levels of self-control.) What’s more, these beliefs had a tangible impact on their college grades, as students with a non-limited view of self-control got a significantly higher GPA when taking heavy course loads.

It’s a fascinating paper, limited mostly by its reliance on self-reports. (The GPA data is the lone exception.) I’m not sure how much we should trust a college student’s retrospective summary of his or her self-control performance, or how those reports might be shaped by their implicit beliefs. Are you more likely to notice and remember your failures of willpower if you believe the will is bound to fail? I have no idea. But it would be nice to see future studies track our lapses in a more objective fashion, and hopefully over a longer period of time.

That said, these quibbles obscure a bigger point. We are constantly being besieged with bodily urges that we’re trying to resist. Maybe it’s a rumbling belly, or a twitchy attention, or a leg muscle pooling with lactic acid. Although we know what we’re supposed to do – not eat a candy bar, stay on task, keep working out – it’s hard for the mind to persist. And so we give in, and tell ourselves we didn’t have a choice. The flesh can't be denied.

But here’s the good news: we’re probably tougher than we think. In one paper, published last year in Frontiers in Human Neuroscience, subjects who were flashed happy faces for 16 milliseconds at a time – that’s way too fast for conscious awareness – pedaled a bike at an intense pace for 25 minutes and 19 seconds. Those flashed sad faces only made it for 22 minutes and 22 seconds. (An even bigger boost was observed after some cyclists were primed with action words, such as GO and ENERGY.) What caused the difference? The subliminal faces didn’t strengthen their muscles, or slow down their heart rate, or mute the pain in their quads. Instead, the visuals provided a subtle motivational boost, which helped the cyclists resist the pain for 12 percent longer.

These results suggest, like the recent Job, et al. paper, that our failures of self-control are primarily not about the physical limits of the brain and body. Those limits exist, of course – the legs will eventually give out and the frontal lobes need glucose. But in the course of an ordinary day, those brute limits are far away, which means that the constraining variable on self-control is often psychological, tangled up with our motivations and expectations. And that’s why our implicit beliefs about self-control and the mind can be so important. (And also why we need to ensure our kids are given the most useful beliefs, which Dweck refers to as a "growth mindset."*) If you believe the self is weak or that the mind is fixed – say, if you’ve read all those ego depletion papers – then you might doubt your ability to stay strong; the lapse becomes inevitable. “A nonlimited theory does not turn people into self-control super heroes who never give in to temptations,” write Job, et al. “However, they lean in when demands on self-regulation are high.” The self they believe in does not wilt after choosing a radish; it is not undone by a long day; it can skip the lemonade and still keep it together.

We are not perfect. Not even close. But maybe we’re less bound to our imperfections than we think.

*In a new paper, Paunesku et. al. show that offering high-school students a brief growth mindset intervention - teaching that them "struggle is an opportunity for growth," and not an indicator of failure - led to significantly better grades among those at risk of dropping out. While this result itself isn't new, Paunesku showed that it was scalable, and could be delivered to thousands of students at low-cost using online instruction.

Job, Veronika, et al. "Implicit theories about willpower predict self-regulation and grades in everyday life." Journal of Personality and Social Psychology (April 2015)