On 21 October 2026, the journal Proceedings of the Royal Society B published a peer-reviewed research article titled "Contagion and escalation of rule violations", with first author Lucas Molleman of the University of Amsterdam. The timing matters for context: the study was published this week, but the experiment itself was run earlier, in the week before the 2024 US presidential election. Keeping that distinction clear — publication date versus when the data were collected — is a small habit of academic care that this article will keep practising throughout.
The study, according to the journal's official record, was a pre-registered online experiment with 750 participants recruited from the USA, roughly half of whom self-identified as Republicans and half as Democrats. Participants played a simple task in which a rule marked which choices were allowed, and breaking the rule offered a higher reward. Outright rule-breaking — choosing the highest-payoff option — appeared in 22.7% of initial choices. The Guardian, reporting on the study on 20–21 October 2026, noted that participants showed large degrees of rule-breaking even at the start, with 23% of choices involving the forbidden highest-reward square.
The central finding is what happened after participants watched others act. Observing major violations strongly increased participants' own rule-breaking, and the effect was strongest when the violator was socially close — in this experiment, someone who shared the participant's political group. The Guardian's report adds that in some cases participants went beyond the violations they had observed. The journal article quantifies this: in 4.4% of trials, participants escalated beyond the observed degree of rule-breaking, a pattern involving 33 participants overall. Simulations in the paper suggest that when individuals occasionally escalate like this, even rare escalation can trigger a rapid shift from widespread compliance to collective rule-breaking.
One important caution belongs in any honest summary. Professor Stephen Reicher of the University of St Andrews, who was not involved in the work, told the Guardian that the experiment was abstract and involved no penalty for rule-breaking, so its relevance to real-world examples remained unproven. That caution is worth carrying forward: this study shows a mechanism in a controlled setting, not a measured outcome in classrooms.
Why should students care about a psychology experiment? Because the mechanism it describes — rule-breaking spreading through observation, and spreading fastest through people you feel close to — maps directly onto situations students face every term. You sit next to someone who copies a problem set. Your group chat shares an unauthorized solution file. A classmate submits an essay with unattributed AI-generated text and gets a better grade than you did with your own work. The study suggests these observations are not neutral; watching others break rules measurably pulls observers toward breaking rules themselves, and the pull is strongest from people in your own circle.
So what can you actually do? The following are practical suggestions from this article's editorial perspective — not official rules, not proven interventions, and not guarantees of any outcome.
First, decide your line in advance. The experiment showed that people's choices shifted after seeing others act; a decision made in the moment is more vulnerable to that pull than a decision made beforehand. Before a course starts, work out where you stand on the specific rules that apply: what counts as acceptable collaboration, how you will cite sources, what your institution permits regarding AI tools, and what individual work means in each class. Write it down if it helps. A line decided calmly is easier to hold than one improvised under pressure.
Second, separate your decision from what others do. The study's core result is precisely that observation changes behaviour. That is a reason to notice the influence, not a reason to obey it. When you catch yourself thinking "they did it, so why shouldn't I", treat that thought as evidence that the contagion effect is operating on you — and as a cue to return to the line you set in advance.
Third, treat "everyone does it" as a risk signal rather than a justification. The claim that everybody cheats is usually unverified, and even if rule-breaking were common, the study suggests visibility is what spreads it. Hearing that everyone does it is itself an observation of rule-breaking, with the same pull the experiment measured. Let the phrase alert you rather than persuade you.
Fourth, take the "socially close" finding seriously in both directions. Influence was strongest from people participants identified with — their own political group in the study. For students, the equivalent closeness is often the study group, the friendship circle, the team. This cuts two ways. It means the people most likely to pull you off your line are your friends, and it also means your friends are the people you can most effectively agree boundaries with. A short, honest conversation before a group project — "I'm keeping this individual, and here's how I'll handle sources" — costs little and sets a norm the whole group can follow. Norms agreed among peers are often sturdier than norms imposed from above.
Fifth, anchor yourself in your institution's official rules rather than in classroom folklore. What is allowed varies by course, instructor, and institution, especially around collaboration and AI tools. When you are unsure, check the official syllabus, academic integrity policy, or ask the instructor directly. The experiment's rule was arbitrary; yours are written down, and following the written rule removes the ambiguity that contagion feeds on.
Finally, remember that the study itself contains an encouraging fact. Even in a setting with no penalties, where rule-breaking carried a clear reward, a substantial share of participants consistently followed the rules — the Guardian reports about a third did so even when it brought no largest reward. Rule-following under social pressure is not naive; it is common, and the paper's simulations show that collective outcomes depend on whether escalation stays rare or spreads. Your individual choice is one data point in that collective picture, and holding your line makes it easier for others nearby to hold theirs.
A note on limits, in the spirit of Reicher's caution: this experiment was abstract and penalty-free, so nobody should claim it proves what happens in real classrooms. What it offers is a plausible mechanism, consistent with everyday experience, that gives students a reason to build defences in advance.
Here is a concrete next action for this week: pick one course you are taking, find its official rules on collaboration, citation, and AI use, and write down in two or three sentences where your own line sits — then share it with one study partner and ask where their line sits. That single conversation turns an abstract finding into a practical norm between two people, which is exactly the scale where the study says influence matters most.
中文摘要
2026年10月21日,《皇家学会B辑》发表同行评审研究《规则违反的传染与升级》(Molleman等人)。这项预先注册的实验招募了750名美国参与者,发现观察他人严重违规会显著增加自身的违规行为,当违规者与自己属于同一政治群体时效果最强;在4.4%的试验中,参与者甚至超出所观察到的违规程度,模拟显示罕见的升级即可使群体从普遍守规迅速转向集体违规。《卫报》在2026年10月20日至21日的报道中记录了圣安德鲁斯大学Stephen Reicher教授的提醒:该实验抽象且无惩罚,对现实世界的适用性尚未被证明。本文据此为学生提出编辑建议:提前确定自己的底线;把自己的决定与他人的行为分开;把"大家都这样"视为风险信号而非理由;由于亲近者影响最大,应与学习小组的朋友事先约定边界;以学校官方规则为准。这些仅为编辑建议,并非官方规定或结果保证。建议本周采取一个具体行动:选定一门课程,查清其关于合作、引用和AI使用的官方规则,写下自己的底线,并与一位学习伙伴分享交流。