Thèse Évolution Culturelle Artificielle des Croyances Expérience de Flexibilité H/F - Doctorat.Gouv.Fr

Établissement : Université Grenoble Alpes École doctorale : MSTII - Mathématiques, Sciences et technologies de l'information, Informatique Laboratoire de recherche : Laboratoire d'Informatique de Grenoble Direction de la thèse : Jérôme EUZENAT ORCID 0000000178084046 Début de la thèse : 2026-10-01 Date limite de candidature : 2026-07-31T23:59:59 Cultural evolution is the application of evolution theory to culture. It has been applied to various aspects of our life in societies: from customs to languages, from boat shapes to company structures [Messoudi, 2011]. In our context, culture is the beliefs and knowledge of agents, that determine their behaviour. Cultural evolution has been the subject of multi-agent simulation [Axelrod, 1997; Steels, 2012; Acerbi et al., 2022]. Artificial cultural evolution, like artificial intelligence for intelligence, aims at considering the general principles and mechanisms governing cultural evolution.For that purpose, we aim at defining a model of cultural evolution experiments that allows different types of agents to play different types of games. This model will be supported by a simulation environment to test cultural evolution hypotheses and ensure the reproducibility and availability of such experiments. We also seek at promoting this approach towards social scientists interested in cultural evolution.

In this context, this PhD proposal aims at instantiating this general model into specific experimental designs investigating two specific directions.

The first direction is to reproduce and extend previous experiments on knowledge and belief evolution [Bourahla et al., 2021; 2022], within a more flexible simulation framework. These experiments will study the influence of multiple populations and generations of agents on the quality and diversity of knowledge developed while playing different games. This will require a conceptual reflection on the nature of populations and generations.

The second direction aims at identifying in such experiments the source of knowledge variation, transmission and selection so that they can be controlled. Indeed, within cultural evolution, they may occur at different stages, e.g. vertical and horizontal transmission, selection by the environment and selection by the agents, and may be combined, e.g. variation or selection occurring at transmission time. Moreover, an occurrence may compensate another: variation during vertical transmission may be compensated by variation during horizontal transmission [Bourahla et al., 2022]. Experiment designs should make it explicit in order to determine that these three operations are indeed necessary for knowledge to evolve, and to characterise their relative influence on the quality and diversity of the resulting knowledge.

These experiments will directly inform the general model design.
Cultural evolution is the application of evolutionary theory to culture. Like evolution it relies on variation, transmission and selection. However, these can occur in various forms which can compensate each other. Multi-agent simulations can be used to understand how this happens and how it affects agents' culture.
The recruited candidate will have for main tasks to:
- design experiment settings, and
- develop simulations regarding these two axes,
but other tasks include:
- participating to the definition of the general model,
- contributing to the development of the simulator, and
- taking part to the ACBE project dissemination effort, in particular through publishing papers.

- Curiosity and openness.
- Interaction with other researchers.
- Autonomous researcher.
- Interests in agent-based simulation and/or cultural evolution.
- Innovative.