Ongoing Research

The ability to form judgments about one’s own cognitive performance—known as metacognition—has been widely studied across domains. More recently, a few studies have shown that individuals can also evaluate the quality of their own metacognition, an ability referred to as meta-metacognition. Yet despite its promising implications, meta-metacognition remains a poorly defined construct. Here, we contribute to this emerging theme with a novel experimental paradigm that offers a precise definition and incentivization of meta-metacognitive judgments, enables direct comparisons between metacognitive and meta-metacognitive levels, and does so using observable data alone—without relying on unobserved model parameters. In a first experiment, participants judged which of two perceptual decisions was more likely to be correct (a metacognitive decision), and then rated their confidence in this judgment (a meta-metacognitive evaluation). In a second experiment, we incorporated trial-by-trial probabilistic confidence ratings and a counterfactual urn task to disentangle the information available at the metacognitive and meta-metacognitive levels. Crucially, we show that meta-metacognitive judgments predict the accuracy of metacognitive decisions beyond what is captured by participants’ probabilistic confidence ratings alone. Moreover, we identify systematic inconsistencies between confidence ratings and metacognitive decisions—inconsistencies that improve decision accuracy—revealing that metacognitive evaluations draw on multiple, dissociable sources of information. Together, these findings provide direct evidence for meta-metacognitive insight, while challenging the assumption that the metacognitive level is functionally unified.

Women of similar ability are less likely than men to choose competitive compensation schemes, a gap with important implications for career advancement (Niederle & Vesterlund, 2007). Yet many real-life career decisions, such as negotiating a raise or applying for a promotion, are made after discussing them with others, particularly romantic partners. Despite the potential importance of these interactions, little is known about how interpersonal discussions shape individuals' willingness to compete, and whether their influence depends on whom individuals discuss with.

To address these questions, we conduct a preregistered laboratory experiment in which participants attend with their real-life romantic partners and are randomly assigned to discuss an upcoming competitiveness decision either with their partner or with an unfamiliar participant of the opposite sex. Before and after the discussion, one participant in each pair, the decision-maker, chooses how to allocate their compensation for a math task between a piece-rate and a tournament scheme. By varying the identity of the discussion partner and the gender of the decision-maker, we investigate whether discussions lead to larger revisions in competitiveness when they involve romantic partners rather than strangers, and whether women are more responsive to these discussions than men.

We also investigate the mechanisms through which discussions may influence competitiveness. In particular, we examine whether changes in willingness to compete are associated with revisions in self-confidence about one's performance or with the discussion partner's own preferences regarding competition. By combining measures of competitiveness, confidence, risk attitudes, and partners' preferences with recordings of the discussions, we aim to better understand how interpersonal communication shapes competitive decisions and whether its influence differs across genders and relationship types.

Many economic and social decisions depend on individuals' beliefs about others' preferences, intentions, and likely behavior. Whether deciding whom to trust, how to cooperate, or how to resolve disagreements, accurately understanding others can be crucial for making informed decisions. Yet little is known about how accurately individuals form such beliefs, how confident they are in their knowledge, and how they acquire information about others' preferences. One important channel is interpersonal communication: in everyday life, individuals often learn about others not by directly observing their choices, but through the conversations they have with them. These questions are particularly relevant within romantic relationships, where partners interact regularly and make important joint decisions, such as financial investments or housing choices, that depend on their understanding of each other's preferences.

To address these questions, we conduct a preregistered laboratory experiment in which participants attend with their real-life romantic partners and are randomly assigned to engage in a live discussion either with their partner or with an unfamiliar participant of the opposite sex. Participants answer 15 binary-choice questions covering everyday preferences, hypothetical economic decisions, and incentivized choices drawn from standard experimental economics paradigms. They then predict how both their real-life partner and another participant in the laboratory answered these questions, reporting their confidence in each prediction.

We investigate how accurately individuals predict their romantic partners' preferences and choices, and whether their confidence in these predictions reflects their actual knowledge. We then examine whether a brief discussion with an unfamiliar person improves prediction accuracy and confidence, and how the resulting knowledge compares with that developed within a romantic relationship. Finally, we compare human predictions with those generated by large language models provided with discussion transcripts, to better understand how information about others' preferences is extracted from interpersonal communication. Together, these analyses shed light on how well individuals understand their partners and the extent to which discussion can help them form more accurate beliefs about others.

Many important economic decisions, such as buying a home, planning a move, or making financial investments, require individuals to interpret and combine information that is often dispersed across different sources. Within households, partners may possess different pieces of relevant information, making communication essential for informed decision-making. Yet recent research suggests that information exchange within couples is often imperfect, even when partners have aligned incentives. In particular, some studies document gender differences in how individuals incorporate information provided by their partners (Conlon et al., 2021; Fehr et al., 2024), although these findings have not always been replicated (Mustafi, 2024). Whether such differences reflect how information is communicated or how it is received remains unclear.

To investigate these questions, we conduct a preregistered laboratory experiment in which participants attend with their real-life romantic partners and are randomly assigned to discuss either with their partner or with an unfamiliar participant of the opposite sex. We study information transmission in two domains: individuals' relative position in the national income distribution and past housing-price inflation. In each domain, one participant receives accurate information while the other does not. We elicit participants' beliefs before and after the information exchange, allowing us to measure whether discussion improves the accuracy of the uninformed participant's estimates. We also collect measures of confidence, expectations about future housing prices, and an incentivized investment decision involving a housing-market fund.

We investigate whether discussion facilitates the transmission of information, whether women benefit more from these exchanges than men, and whether communication is more effective between romantic partners than between strangers. We further examine how confidence and features of the discussions relate to belief updating, and whether the identity of the discussion partner influences inflation expectations and investment decisions. Together, these analyses aim to shed light on how individuals combine dispersed information through interpersonal communication and whether romantic relationships facilitate or hinder this process.

Gender norms play an important role in shaping the division of responsibilities and decision-making within households. Yet individuals may be influenced by different normative expectations: their own beliefs about what is appropriate, the norms they perceive as shared within their couple, and those prevailing in society more broadly. These different norms may not always coincide, raising questions about which are most influential in shaping household decisions. Moreover, while previous research has documented important differences in gender-role attitudes and household specialization across countries, much of this evidence relies on self-reported attitudes, leaving open the question of how different normative expectations translate into actual behavior.

To address these questions, we propose a laboratory experiment with real-life romantic couples in France and Germany. Using incentivized coordination games, we elicit three distinct types of gender-role norms: individual norms, couple norms, and perceived societal norms. Participants evaluate a series of statements and scenarios concerning gender roles in professional and family life, allowing us to examine whether couples develop shared normative expectations that differ from both partners' personal beliefs and broader societal norms. We then measure participants' willingness to assume the role of household head in an incentivized decision that determines who will make subsequent decisions affecting the couple's earnings.

We investigate which normative expectations best predict household decisions, whether couples possess distinct shared norms, and how these relationships differ between France and Germany. We also aim to examine whether a brief intervention can shift gender-role norms, which normative expectations are most responsive, and whether these changes translate into actual behavior. Together, these analyses seek to better understand how individual, couple, and societal norms interact in shaping household decisions and whether targeted interventions can influence these dynamics.

This project investigates whether the quality of user interfaces (UI) in economic experiments affects the reliability of behavioral data. While experimental economists devote significant attention to designing robust incentives and clear instructions, comparatively little scrutiny has been given to the interfaces through which participants interact with experimental tasks.  Drawing on insights from web design and human-computer interaction, we test whether UI flaws—ranging from poor visual design to inconsistent input methods—can degrade data quality by increasing noise, reducing engagement, or inducing bias.

In our experiment, participants complete three widely used experimental tasks—a Bayesian updating task, a real-effort task, and a risk preference elicitation task—under either a well-designed or a poorly designed interface condition. The two versions differ along several dimensions: text presentation (e.g., inconsistent fonts and typos), visual layout (e.g., misaligned elements and clashing colors), input methods (e.g., inconsistent use of keyboard and mouse), and responsiveness (e.g., unnecessary confirmation steps). In addition, this design allows us to examine both involuntary effects (e.g., distraction or cognitive load) and voluntary reactions (e.g., reduced effort or strategic disengagement).

The project aims to inform best practices for experimental design in economics and behavioral sciences, particularly in online settings where UI issues are both more likely and more consequential. By identifying the behavioral impact of seemingly superficial design flaws, we seek to promote greater attention to interface quality as a key determinant of data validity.