We test experimentally the impact of reducing competitors’ chances of winning (sabotage) on cooperation. Participants compete with or without the sabotage option and play a public good game before and after the exposure to a competitive environment. Contributions to the public good decrease by 9% of the initial endowment when observing pre-post cooperation decisions. Nonetheless, we find no evidence that competition with sabotage further reduces contributions to the public good. Yet, the amount of sabotage received negatively predicts contribution within treated participants.

We study the impact of donation programs on sustaining cooperation after exposure to a competitive environment. We replicate experimental conditions to study the impact of tournament incentives on willingness to cooperate found in Buser and Dreber (2016) and introduce variations in which competition winners can publicly donate to a charity. Our analysis informs on the effectiveness of donation programs in mitigating the potential negative spillover effects of competition on cooperation. We find that competition does not affect cooperation, resulting in the replication being unsuccessful. Furthermore, donation decisions made after a competition reflect winners' prosocial behavior, but they do not change the cooperation of competition losers.

Empirical Study: 

We study whether structured reflection on incremental progress can reduce preventive disengagement in a cognitively demanding task. Using a laboratory experiment with 400 (planned) university students, we randomly assign participants to one of four between-subject conditions: a Progress Reflection diary (PR), in which prompts guide participants to identify improvements; a Difficulty Reflection diary (DR), in which prompts direct attention to obstacles and failures; a Non-Directional Reflection diary (ND), eliciting open-ended task observations without an evaluative frame; or a Neutral Transcription control (NT), in which participants copy a task-irrelevant text. In each of 15 incentivized rounds, participants choose between attempting a visuospatial sequence-recall task or taking a safe outside option, allowing us to directly measure preventive disengagement. We hypothesize that prompting participants to reflect on small wins reduces skipping and improves performance, while focusing on difficulties amplifies disengagement. This design contributes to the literature on brief psychological interventions in education by isolating the causal

role of progress framing in sustaining engagement, with implications for the design of low-cost, scalable motivational tools. Data collection is ongoing, with results expected by June 2026.

Experimental Study: Data Collection

Empirical Study: 

This study investigates the relationship between misinformation and its emotional content, focusing on three dimensions: emotional intensity (overall emotional tone of the content), emotional valence (positive vs. negative emotions), and the emotional intensity associated to specific emotions (anger, fear, sadness, happiness, surprise, and disgust). Leveraging a novel classification method based on OpenAI's pretrained language models, we generate emotion scores for news content from both the perspective of an impartial media assistant and through the lens of Democrat and Republican voters. Our analysis reveals that emotional content is positively associated with misinformation, highlighting the importance of all emotions, but sadness. Furthermore, we observe heterogeneous patterns across simulated political perspectives, with Republican-leaning scores generally showing stronger emotional responses to misinformation than Democrat-leaning scores.

Experimental Study: Data Collection