Abstract Teamwork is prevalent in many forms of production, but individual contributions to team output are often ambiguous. We use a large-scale online experiment with an interactive team task to investigate gender differences in claiming credit for contributions to teamwork with misaligned incentives: Team members have an incentive to exaggerate their contribution, but this harms the other team member. In this setting, we observe that men claim to have contributed more than equally contributing women. Using two further between-subject treatments, we find no evidence that gender differences in altruism or in overconfidence explain the observed gender claim-gap. Further, the gender gap is present only among high contributors, and we show that the gap is particularly pronounced for large claims, which has potentially important consequences for gender equity in labour market outcomes.
Abstract We study whether and why the mode of decision-making affects reciprocation in a binary trust game, using three preregistered online experiments. Trustees – some of whom were exogenously induced to be surprised by trust – either decided whether to reciprocate after observing their partner’s choice ( hot mode) or committed in advance to a contingent response plan ( cold mode). In our Baseline experiment, once participants gained experience, reciprocation became more likely under hot than under cold decision-making, especially when trust was unlikely – consistent with a model in which positive surprise at being trusted increases the utility of reciprocating. Results from the other two experiments show that surprise promotes reciprocation only when trust reflects an intentional choice by the trustor and when cognitive load in the decision environment is not too high.
Abstract We introduce a general test of capacity-constrained learning models. Learning has capacity constraints when the set of possible ways to learn is exogenously fixed, as in the widely used fixed-capacity versions of rational inattention (Sims, 2003) and efficient coding (Woodford, 2012). With such models, changes in incentives do not alter the extent of attention, only how individuals decide to allocate their scarce attention. We show that choice data are consistent with capacity-constrained learning if and only if they satisfy a No Improving (Action or Attention) Switches (NIS) condition. Based on existing experiments in which the incentives for being correct are varied, we find strong evidence that participants fail NIS for a wide range of standard perceptual tasks: identifying the proportion of ball colors, recognizing shapes, and counting the number of balls. We further show that violations of NIS occur systematically in response to higher incentives, suggesting that incentives often expand attention beyond what capacity-constrained models allow. However, we find that this is not true for all existing perceptual tasks in the literature, which offers insights into settings where we do or do not expect incentives to impact the extent of attention.
Abstract We study how large language model (LLM) based recommendations affect evaluations of uncertainty. We elicit certainty equivalents for (i) lotteries that share the same reduced-form winning probability but vary in cognitive demands because they are compound, and (ii) an Ellsberg-style ambiguous lottery. Across all lottery types, valuations from participants who receive LLM-based recommendations before each valuation exhibit second-order stochastic dominance relative to valuations from participants who do not. Hence, LLM-based recommendations leave mean valuations unchanged but compress the distribution, reducing the incidence of extreme valuations. Consistent with this pattern, LLM-based recommendations reduce the intensity of risk, complexity, and ambiguity aversion, while leaving the fraction of participants classified as averse unchanged.
Groups often agree that change is needed, but differing preferences and information may still lead to difficulty coordinating on a particular alternative. In a laboratory experiment, we study coordination and information aggregation in a divided-majority voting game with public information and conflicting preferences. Voters observe private signals and a public signal about whether alternative upper A $A$ A or upper B $B$ B is socially optimal. Unless sufficiently many voters agree, an inferior default upper C $C$ C occurs. When the public signal is more accurate than the private signal and voters have common preferences, most voters coordinate on the public signal. However, when public-signal accuracy is reduced or preferences conflict, voters follow the public signal less frequently. While reduced public-signal accuracy substantially increases the frequency of coordination failure, conflicting preferences between subgroups do not. These results highlight the importance of accurate public information in achieving collective action.
