Abstract: Using a real-effort experiment, we study whether group identity affects unethical behavior in a contest game. We vary whether minimal group identity is induced or not, whether individuals have to report their own outcome or the outcome of their competitor, and whether pairs of competitors share the same group identity or not. We show that individuals misreport in the same proportion and to the same extent by inflating their outcome or by decreasing their opponent’s outcome, except when any possible scrutiny by the experimenter is removed. Regardless of the possibility of scrutiny by the experimenter, misreporting is affected neither by the competitor’s group identity nor by the individual’s beliefs about others’ misreporting behavior. This suggests that in competitive settings, unethical behavior is mainly driven by an unconditional desire to win.

Abstract: We experimentally test whether the gap between reference and actual income impacts subsequent altruism. Participants first perform a real-effort task for a fixed wage and then play a dictator game. Between conditions, we vary the level and the timing of the revelation of the wage. In some conditions, participants know the wage before the real effort task and are not informed of the other potential levels. In some other conditions, they are informed of the distribution of wages before the real effort task, but the actual wage is only revealed afterward. Participants in the latter conditions can form references that may be higher or lower than their actual wage. Our hypothesis is that the gap between the reference and the actual wage impacts transfers in the subsequent dictator game, either because participants want to compensate their recent losses, or because of the emotional reaction to gains and losses. The results support this hypothesis: participants who get the low wage transfer less and are less likely to transfer when they are informed of the other potential levels than when they are not. Conversely, participants who get the high wage are more likely to transfer positive amounts when they are informed of the other potential levels. We use physiological (skin conductance response) and declarative data to discuss the role of emotions in our treatment effects. 

Abstract: The neural circuitry involved in moral decisions has been studied since the early days of cognitive neuroscience, mainly using moral dilemma. However, the neurocomputational mechanisms describing how the human brain makes moral decisions and learns in various moral contexts are only starting to be established. Here we review recent results from an emerging field using model-based fMRI, which describes moral choices at a mechanistic level. These findings unify the field of moral decision making, extend a conceptual framework previously developed for value-based decision making and characterize how moral processes are computed in the brain. Moral dilemma can be modeled as value-based decisions that weigh self-interests against moral costs/harm to others and different types of prediction errors can be distinguished in different aspects of moral learning. These key computational signals help to describe moral choices and moral learning at an algorithmic level and to reveal how these cognitive operations are implemented in the brain. These researches provide a foundation to account for the neurocomputational mechanisms underlying moral decision making. 

Abstract: Social discounting predicts that one’s concern for others decreases with increasing social distance. Cultural dimensions may influence this social behavior. Here, we used a dictator game, in which the participants and real members of their social entourage profited from the partition of the endowments determined by the participant, to compare how Chinese and French university students shared endowments with people at different social distances. We tested two hypotheses based on the concepts of kinship premium and cultural collectivism. Stronger ties between close family members were expected among Chinese. This may predict a larger “kinship premium,” i.e., increased generosity to family members at close social distances, in Chinese relative to French participants. Similarly, because collectivism is thought to be stronger in Asian than western societies, greater generosity at larger social distances might also be expected among Chinese participants. The results showed that Chinese were more generous than French at close social distances but discounted more as social distance increased. This difference between French and Chinese was confined to family members and no significant difference in generosity was observed between French and Chinese for non-family members at any social distance. Our findings evidence a stronger kinship premium among Chinese than French students, and no significant effect of cultural collectivism. 

Abstract: Costly punishment of social norms transgressors by third-parties has been considered as a decisive stage in the evolution of human cooperation. An important facet of social relationship knowledge concerns the strength of the social ties between individuals, as measured by social distance. Yet, it is unclear how the enforcement of social norms is influenced by the social distance between a thirdparty and a norm violator at the behavioral and the brain system levels. Here, we investigated how social distance between punishers and norm-violators influences third-party punishment. Participants as third-party punished norm violators more severely as social distance between them increased. Using model-based fMRI, we disentangled key computations contributing to third-party punishment: inequity aversion, social distance between participant and norm violator and integration of the cost to punish with these signals. Inequity aversion increased activity in the anterior cingulate cortex and bilateral insula, and processing social distance engaged a bilateral fronto-parietal cortex brain network. These two brain signals and the cost to punish were integrated in a subjective value signal of sanctions that modulated activity in the ventromedial prefrontal cortex. Together, our results reveal the neurocomputational underpinnings of third-party punishment and how social distance modulates enforcement of social norms in humans.

Abstract: Observing immoral behavior increases one’s dishonesty by social influence and learning processes. The neurocomputational mechanisms underlying such moral contagion remain unclear. We tested different mechanistic hypotheses to account for moral contagion. We used model-based fMRI and a new cheating game in which participants were sequentially placed in honest and dishonest social norm contexts. Participants’ cheating behavior increased in the dishonest norm context but was unchanged in the honest. The best model to account for behavior indicated that participants’ valuation was dynamically biased by learning that others had cheated. At the time of choice, the internalization of social norms was implemented in the lateral prefrontal cortex and biased valuations of cheating. During learning, simulation of others’ cheating behavior was encoded in the posterior superior temporal sulcus. Together, these findings provide a mechanistic understanding of how learning about others’ dishonesty biases individuals’ valuation of cheating but does not alter one’s established preferences.

Preliminary abstract: Gambling disorders is often associated with an impairment of cognitive abilities associated with gambling. Yet, gambling decisions are known to be a source of many different biases. Here, we ask how these biases are affected in gambling disorders ? We focus on the Gambler's Fallacy (GF) which happens when individuals are facing a set of consecutive identical outcomes leading them to predict that the alternative outcome likelihood increases. We designed a color betting roulette wheel game, and compared behavior of a sample of pathological gamblers to a healthy control sample. Results show that pathological gamblers GF-related behavior are affected by a previous winning bet while control participants are not. Using fMRI, we related this behavioral difference to an increased activity in the dorso-lateral Prefrontal Cortex (dlPFC) and the dorsal anterior cingulate cortex after a previous win in pathological gamblers compared to control participants. Pathological gamblers and control participants behavior is explained by the same computational model in line with the lack of difference in GF between the two sets of participants. Despite this equivalent, we found that the bilateral dlPFC is encoding both value of the chosen option and the GF option but only for pathological gamblers. Overall our results are in line with the idea that gambling disorder is associated with an impairment of cognitive functions even if this does not reflect into behavioral differences between pathological gamblers and control participants.

Abstract: Social networks play a crucial role in facilitating knowledge exchange, serving as a primary avenue for social learning. In today's interconnected world, where misinformation abounds, creating more informed, harmonious societies requires understanding how individuals gather and interpret social information within their networks. One of the most successful models of social learning in network, DeGroot (DG) model, posits that we adjust our beliefs using the average opinions of individuals in our social network. In contrast, recent works have proposed that reinforcement learning (RL) models can better capture learning in social networks. These classical models in decision neurosciences assumes that learning is accomplished by sequential updating based on the difference between expectation and choice outcomes. This raises the question of how these two very different types of learning capture the propagation of information in social network. To address this question, we used these models to fit data from two recent studies on propagation of information in social networks. We found that social learning behavior can be better captured with the DG model, despite systematic variations in adopted learning strategy across tasks and network architectures. In addition, through model simulations, we examined similarities and differences between two models and how they can be detected in future experiments. Overall, our results provide new insights into the mechanisms of information aggregation in social network and their underlying computational mechanisms.

Abstract: We study the impact of competition on morals using a dynamic variant of the die-under-the-cup task. Players can repeatedly misreport the outputs of consecutive die rolls to earn more money, either under an individual piece-rate pay scheme or in a two-player tournament. In this dynamic setting, we disentangle the effect of the incentive scheme and the effect of information provision about one’s relative performance, by comparing settings with continuous vs. final ex post feedback on the counterpart’s reported outcome. We find that individuals lie more under competitive rather than non-competitive incentive schemes, but only if both players can cheat in the tournament. Continuous feedback on the counterpart’s reports does not increase cheating in the tournament, while it does under the piece-rate scheme. These findings shed light on the effects that different competition and information policies have on morals in occupational settings.