Why do some people keep making the same harmful choices, even when they know better? A global study has revealed three distinct decision-making types and why punishment doesn’t work for everyone.
When someone engages in behaviors that actively undermine their own goals, success, or well-being, that’s known as self-sabotage. Sometimes conscious and sometimes unconscious, self-sabotage can manifest in various areas of life, including work, relationships, and personal achievements.
A new study by the University of New South Wales (UNSW), Sydney, examined self-sabotage, wanting to understand why some people continue to make choices that hurt themselves or others, even when they’ve been punished for it. Instead of assuming everyone responds to punishment the same way, the researchers tested whether people actually learn differently from punishment.
“We found that some people just don’t learn from experience,” said the study’s corresponding author, Dr Philip Jean-Richard Dit Bressel, a behavioral neuroscientist and experimental psychologist at UNSW’s School of Psychology. “Even when they’re motivated to avoid harm and are paying attention, they fail to realize their own behavior is causing the problem.”
The researchers recruited 267 individuals from 24 countries, representing a diverse range of ages and backgrounds. Participants were asked to play a computer game with the goal of earning as many points as possible. Clicking on one of two planets would reward points. One was a “risky” planet that sometimes triggered an attack, resulting in a loss of points; the other was a “safe” planet that never led to a loss of points. After three punishment rounds, participants were explicitly told which planet triggered attacks, and their understanding was confirmed via a test.
“We basically told them, ‘this action leads to that negative consequence, and this other one is safe’,” Jean-Richard Dit Bressel said. “Most people who had been making poor choices changed their behavior immediately. But some didn’t.”
Data collection took place in two stages, comprising the initial test and a retest six months later. At each stage, participants played the computer game and completed self-report questionnaires that measured their gaming strategy, cognitive flexibility, habitual tendencies, and alcohol use. Of the participants who had been recruited, 128 returned for follow-up testing six months later. The researchers found that participants fell into three distinct behavioral types:
- Sensitive. These people quickly recognized which planet caused harm and changed their behavior early, even before they were told.
- Unaware. They failed to infer the punishment cause at first, but once it was explicitly explained to them, they changed their behavior.
- Compulsive. This group didn’t change their behavior even when the punishment had been explained to them. They continued choosing the harmful option.
The findings mirrored those of a previous study, which involved only Australian psychology students, rather than the diverse international cohort used in the new study. In the prior study, about 35% were Sensitives, 41% were categorized as Unaware, and 23% fell into the Compulsive category. In the present study, about 26% were Sensitives, 47% were Unawares, and 27% Compulsives.
“We ran the same task with a general population sample from 24 countries – people of different ages, backgrounds, and life experiences,” said Jean-Richard Dit Bressel. “And what we found was that the same behavioral profiles emerged. Everything we’d seen in the Australian psychology students replicated almost exactly.”
Jean-Richard Dit Bressel said that the slight difference in numbers between the studies is partially attributable to the number of over-50s included in the present study, who were more likely to fall into the Compulsive category.
“This may be linked to cognitive flexibility – the ability to adapt your thinking,” he said. “And that tends to decline with age.”
And the behavior of those participants in the Compulsive category couldn’t be explained away as laziness or poor motivation; it was a failure to integrate knowledge into action. The researchers made a point of asking what was behind those choices.
“We asked participants what they thought was the best strategy, and they often described exactly what they were doing – even when it was clearly the wrong choice,” said Jean-Richard Dit Bressel.
Interestingly, most participants remained in the same behavioral type after six months.
“That was one of the more striking findings,” Jean-Richard Dit Bressel said. “It suggests these aren’t just random mistakes or bad days. They’re stable traits – almost like personality types. This is not to say they’re fixed, just that they may require intervention to break.”
The study has some limitations. Primarily, it was conducted in a simplified, game-like environment – real-life behaviors may involve more complex emotions, contexts, and stakes. Second, participants were fluent English speakers with internet access, which may limit the applicability of the results to broader populations.
Nevertheless, the study shows that people don’t always learn from punishment the same way, which could have important real-world implications. It could explain why fines, warnings, or health campaigns don’t always work, or why some people relapse into harmful behaviors like substance use. The study’s findings could inform behavioral therapy and coaching, which could be tailored to each behavioral type. They could also inform the design of more effective interventions in the addiction and criminal justice spaces.
“Of course, real life is far more complex than the simple game we devised,” said Jean-Richard Dit Bressel. “But the patterns we’re seeing, where people ignore both experience and information, are similar to what we see in gambling and other compulsive behaviors. We’ve shown that standard information campaigns work for most people – but not for everyone. For compulsive individuals, we may need a different kind of intervention.”
The study was published in the journal Communications Psychology.
Source: UNSW Sydney
