From

Time

Location LISN Site Belvédère

Interactions with Human, Thesis

Social Performance in Collaborative Problem Solving: Influence of Self-Awareness and Individual Characteristics

Soutenance de thèse, co-dirigée par Jean-Claude Martin, professeur, Céline Clavel, maîtresse de conférences, et Élise Prigent, maîtresse de conférences (LISN, Université Paris-Saclay)

Speaker : Jennifer Hamet

Jury

  • Estelle MICHINOV (Université de Rennes 1, LP3C), Rapporteure
  • Olivier LUMINET (Université catholique de Louvain, IPSY), Rapporteur
  • Alice NORMAND (Université Clermont Auvergne, LAPSCO), Examinatrice
  • Yann COELLO (Université de Lille, SCALab) — Examinateur

Abstract

Social skills, which are socially learned behaviors, enable individuals to adapt to various social situations. Among these, collaborative problem-solving (CPS) skills play an important role in academic and professional success. They are defined as a set of skills for solving problems and progressing toward a common goal while collaborating with group members. We perceive their mobilization through social performance, an indicator of the adequacy of an individual’s responses to the expectations of the situation. This performance results from a dynamic between endogenous (person-related) and exogenous (environment-related) factors. This thesis examines how CPS performance is modulated by social anxiety and alexithymia on the one hand, and by the induction of self-focused attention and the behavior of a virtual partner on the other. In order to investigate this, we built a multimodal corpus of dyads engaged in collaborative tasks with manipulated visual feedback (feedback on oneself, on the other, or combined). Based on this corpus, we validated two French-language scales for assessing social skills specific to CPS. Our results show that self-focused attention can impair social performance, especially in alexithymic individuals; however, in highly anxious participants, performance was preserved or even improved when attention was focused on the self rather than divided between the self and the partner. To isolate the effect of the partner in CPS tasks, we designed virtual agents (high-performing vs. low-performing profiles). Interacting with a high-performing agent increases the user’s social performance, while a low-performing agent decreases it. The results of this second study suggest that these effects are moderated by anxiety and alexithymia. The contributions of this thesis provide new insights into the relationship between endogenous and exogenous factors of social performance in CPS, provide robust measurement tools (SPRS, SSC), a shareable corpus, and open up application prospects via adaptive virtual agents for personalized training of social skills.

Publications

Publications of Jennifer Hamet are available on line: https://hal.science/search/index?q=jennifer+hamet+bagnou