Missions

The successful candidate will conduct a series of experimental studies as part of the CaRe research project (AAP TIRIS – Scaling up), which tests the formalization and study of collective reasoning (i.e., taking into account the mental states and preferences of others) and its role in decision-making in a variety of situations involving interactions between agents.

The successful candidate will conduct experimental studies on situations involving interactions between (a) humans and humans and (b) humans and machines (i.e., conversational agents) in which ethical issues (e.g., social dilemmas, causal reasoning) are involved and for which the preferences and values of the agents involved are not systematically aligned. The aim will be to assess individuals' acceptability, preferences, and perceptions (e.g., perceived warmth, perceived competence, perceived reliability, etc.) of different models based on their ethical and moral values (e.g., fairness, reciprocity, respect for norms, perception of consequences, etc.) and different moderators (e.g., personality traits, etc.). These study situations will focus on interactions between humans and between humans and machines (i.e., large language models).

The successful candidate will work closely with IRIT (Institut de Recherche en Informatique de Toulouse), co-leader of the project, responsible for designing and implementing the artificial agent used in the experimental study of human-machine interaction.

Activités

- Conduct a literature review on the different dimensions of collective reasoning (reasoning about the thoughts of others, team/group reasoning, ethical reasoning).

- Develop experimental protocols for studying collective reasoning:
a. In situations where the participant evaluates the reasoning and behavior of another human (human-human scenario)
b. In situations where the participant evaluates the reasoning and behavior of an artificial agent (human-machine scenario).

- Statistical comparison of human-human and human-machine situations
- Evaluation of the moderating effects of certain participant characteristics: fundamental values (e.g., social dominance), personality traits (e.g., honesty-humility, agreeableness), aversive traits (e.g., psychopathy, Machiavellianism), distinction between self-interest and the interests of others, etc.

Compétences

- Hold a PhD in psychology, cognitive science, or a discipline related to the research topic
- Have solid knowledge in the field of judgment and decision-making
- Know how to conduct a literature review in the field of psychology
- Know how to use the experimental method and develop experimental protocols
- Know how to conduct statistical analyses
- Know how to communicate scientific work at international conferences
- Know how to produce scientific reports
- Know how to collaborate with different partners from other scientific disciplines (computer science)
- Have a good level of professional English
- Show initiative and demonstrate creativity

Interested candidates should apply through the CNRS job portal. Please provide a detailed CV, a cover letter with a summary of the research topics. A letter of recommendation will be appreciated.

Contexte de travail

The post-doctorate will take place in the CLLE laboratory: https://clle.univ-tlse2.fr/ under the supervision of Bastien Trémolière.
The person recruited will work with the project team and the different members of the CaRe project consortium.

Summary of the CaRe project:

The CaRe project is devoted to the interdisciplinary study, from the perspective of artificial intelligence (AI) and psychology, of collective reasoning and its role in decision-making. Collective reasoning is the multi-faceted kind of reasoning of a human or artificial agent that involves reasoning about others' minds, team reasoning and ethical reasoning. In particular, the CaRe project aims to formalize collective reasoning with the help of logic, game theory and social choice theory; develop new algorithms for endowing an artificial conversational agent with collective reasoning capabilities; evaluate the proposed formal models and algorithms through experiments on both human-human (H-H) interaction and human-machine (H-M) interaction. The CaRe project will provide a novel understanding

Contraintes et risques

No risk identified.