Purpose - Collaborative consumption, such as car sharing, speci fi cally implicates customer-to-customer interaction, which must be regulated by service providers (companies, peers and self-regulating communities), comprising different challenges for business organizations. While in conventional business relations, consumers are protected from undesirable customer behavior by laws, regulations (power) in the context of collaborative consumption are rare, so that trust becomes more relevant. It is the purpose of the study to investigate possible mechanisms to prevent undesirable customers in collaborative consumption. Design/methodology/Approach - In between subject designs, samples of 186 and 328 consumers fi lled in experimental online questionnaires with vignettes. Analyses were made of differences among car sharing companies, private persons and car sharing communities in terms of the power of providers, trust in providers and trust in other users of the shared goods, undesirable customer behavior and consumer - provider relations. Findings - Companies, private persons and self-regulating communities differ in terms of perceived power and trust. Participants speci fi cally perceive mainly coercive power with the car sharing company, but with the private person and the community, reason-based trust in other users is perceived as prevalent. Nevertheless, undesirable customer behavior varies only marginally over the models. Originality/value - The present study is the first to investigate measures to prevent undesirable customer behavior over different collaborative consumption models. This enables appropriate identification of market segments and tailoring of services. The study identifies opportunities for companies in contrast to private persons and self-regulating communities and, in doing so, provides important stimulation for marketing strategy and theory development.

Les comportements sociaux coopératifs sont longtemps restés un obstacle aux modèles de choix rationnel, obstacle qu'incarnent des dilemmes sociaux où un individu suivant son intérêt personnel est incité à exploiter la coopération d'autrui à son seul avantage. Je détaillerai tout d'abord comment la coopération peut apparaître un choix sensé lorsque elle est envisagée dans un contexte naturel et réel. Un regard à travers l'anthropologie, la psychologie et la neurobiologie conduit à appréhender la coopération davantage comme une adaptation et un apprentissage que comme un défaut de rationalité. Les émotions jouent un rôle essentiel dans ces processus, et je présenterai en quoi les exprimer aide les êtres humains à se synchroniser et à coopérer. Le sourire est souvent invoqué comme exemple d'un signal universel de coopération et d'approbation, une propriété intimement liée à son expression répétée lors de tâches collaboratives. Malgré tout, on en sait encore peu sur la manière précise dont le sourire et les autres expressions interviennent dans la prise de décision sociale, et en particulier sur le traitement des situations d'incongruence où un sourire accompagnerait une défection. Ce point est le cœur de l'étude expérimentale que je rapporte dans ce manuscrit. J'ai réalisé deux expériences confrontant les participants à un dilemme social dans lequel ils pouvaient investir une somme d'argent auprès de différents joueurs informatisés susceptibles de se l'accaparer, ou, au contraire, de la rétribuer avec intérêts. Les joueurs virtuels étaient personnalisés par un visage dont l'expression pouvait changer après le choix du participant: certains affichaient ainsi des émotions incongruentes avec leur ``décision'' subséquente de rétribuer ou non l'investissement du sujet. Malgré les différences méthodologiques, ces deux expériences ont montré que les expressions incongruentes altéraient la capacité des participants à jauger la propension des joueurs virtuels à rétribuer leurs investissements après une ou plusieurs interactions. Cet effet s'est manifesté tant au travers de rapports explicites que dans les investissements effectués. Dans leurs détails, les résultats de ces expériences ouvrent de nombreuses perspectives expérimentales, et appellent à la construction d'un modèle unifié de la décision sociale face-à-face qui intégrerait les nombreuses connaissances apportées ces dernières années par l'étude des grandes fonctions cognitives, tant au niveau expérimental, théorique que neurobiologique. / For few decades, rational choice theories failed to properly account for cooperative behaviors. This was illustrated by social dilemmas, games where a self-motivated individual will be tempted to exploit others' cooperative behavior, harming them for his own personal profit. I will first detail how cooperation may rise as a reasonable --- if not rational --- behavior, provided that we consider social interactions in a more realistic context that rational choice theories initially did. From anthropology to neurobiology, cooperation is understood as an efficient adaptation to this natural environment rather than a quirky, self-defeating behavior. Because pertinent information is often lacking or overwhelming, too complex or ambiguous to deal with, it is essential to communicate, to share, and to trust others. Emotions, and their expression, are a cornerstone of humans' natural and effortless navigation in their social environment. Smiles for instance are universally known as a signal of satisfaction, approbation and cooperation. Like other emotional expressions, they are automatically and preferentially treated. They elicit trust and cooperative behaviors in observers, and are ubiquitous in successful collaborative interactions. Beside that however, few is known about how others' expressions are integrated into decision making. That was the focus of the experimental study I relate in this manuscript. More specifically, I investigated how decisions in a trust-based social dilemma are influenced by smiles which are either displayed along a cooperative or defective behavior (``congruently'' and ``incongruently'', resp.). I carried out two experiments where participants played an investment game with different computerized virtual partners playing the role of trustees. Virtual trustees, which were personalised with a facial avatar, could either take and keep participants investment, or reciprocate it with interests. Moreover, they also displayed facial reactions, that were either congruent or incongruent with their computerized ``decision'' to reciprocate or not. Even if the two experiments presented some methodological differences, they were coherent in that they both showed that participants were altered in remembering a virtual trustee's behavior if the latter expressed incongruent emotions. This was observed from participants' investments in game, and from their post-experimental explicit reports. If many improvements to my experimental approach remain to be done, I think it already completes the existing literature with original results. Many interesting perspectives are left open, which appeal for a deeper investigation of face-to-face decision making. I think it constitutes a theoretical and practical necessity, for which researchers will be required to unify the wide knowledge of the major cognitive functions which was gathered over the last decades.