If someone’s about to go into a cold swimming pool, they’ll probably use one of two tactics. They might dip a toe in, wade in to the ankles, and slowly, slowly inch their body into the water until they’re completely submerged. Or they’ll just cannonball in, and get it over with.

If it’s not a cold swimming pool someone is entering, but rather the icy waters of corruption, which of these two strategies will they choose? Many would say the first; corruption is often characterized as a “slippery slope,” something into which a person or organization slowly descends as more and more small immoral acts add up. But a new study published in Psychological Science argues that people are more likely to abruptly do something extremely unethical than to slowly build up to it—to cannonball into corruption.

In the study, participants played an auction-style game. Bidders took the role of CEOs of construction companies, while a person handing out money (real or fake, depending on the experiment) acted as a public official. Some of the players got the opportunity to bribe the public official to gain an advantage. In one condition, they could bribe the official immediately with a private vacation, gaining an advantage right from the start for all rounds of play. In the other condition, they could bribe the official in two stages—first with an invite to a banquet and later with the big vacation—which gave them the option of building up to a big advantage over time or sticking to a more modest one.

In a final experiment, not only did they win actual money, but if they chose to bribe the official, it would take money away from one of the other players. In all the conditions, people were more likely to heavily bribe the official right away, rather than working up to it. The golden opportunity for a huge reward (at an unethical price) was too good to give up.

In post-game surveys, bribers tended to rate their behavior as corrupt. So they knew it was wrong, but they did it anyway. That likely created some cognitive dissonance; most people believe they are moral, so if they do something unethical, they have to reconcile that with their self-identity.

Some research does suggest that this sort of reconciliation is easier if people gradually slide down a slippery slope into unethical behavior. It makes sense that it would be less difficult to rationalize a small ethical breach than a big one. As the breaches build up, people’s moral standards may lower over time so that they don’t even see the behavior as unethical anymore. This is also how corruption gets normalized in organizations, other studies say.

But it’s also possible that a one-time transgression might be easier to justify, psychologically, than repeated corrupt acts, “and thus could cause less tension between being a moral person, on the one hand, and enjoying the benefits of dishonesty, on the other hand,” the Psychological Science study’s authors write. Someone might write their corruption cannonball off as a one-time thing that’s not truly indicative of who they are as a person. Under the slippery slope condition, there would have to be repeated reckonings.

“People have an amazing ability to justify dubious behavior,” Paul A.M. Van Lange, an author of the study and a professor at VU University in Amsterdam, told me in an email. “We have latitude in viewing our own actions in terms or morality—and we can often add additional ‘dimensions’ to justify our actions. Practicality is key among them. Or the self-serving argument: Most others would do exactly the same.”

In this study, in the slippery slope condition, the two opportunities to bribe came one right after the other. People might behave differently if their unethical acts were more spaced out. Also there were no punishments for the bribery (except, perhaps, guilt), and “how severe corruption emerges under varying punishment regimes is still unclear,” the study reads.

Danila Serra, a professor of economics at Southern Methodist University who has studied corruption, notes that “the social and cultural dimension of bribery” is also missing from this study. “In a country where corruption is widespread, so that small acts of bribery go unpunished and are actually socially accepted, people may be very likely to engage in low-level bribery,” she told me in an email. “Would this make people more likely to engage in large acts of corruption, if they got the chance? The study seems to answer ‘no’ and suggests that people prefer to engage in one-time large corrupt acts than multiple small acts of corruption. I am not convinced.”

In a study she did, which used a similar bribery game to the Psychological Science study, people were less likely to bribe if they knew they’d be socially judged, but only if they identified with the culture of a country with low corruption. If they identified with higher corruption countries, social judgment did not deter bribery.

Van Lange noted this shortcoming as well. “The slippery slope may often be a social process, not only an individual process,” he wrote.

Both Serra and Van Lange said that more research would be necessary to untangle these things. And it may be more prudent for people who don’t want to get caught to avoid the temptation of seizing a golden opportunity; a 2009 study shows that watchdogs are more likely to accept corruption when it adds up slowly than if there’s a huge shift in someone’s behavior all at once. A cannonball, after all, makes a huge splash.