Postdoctoral position (M/F) – Formalizing ambiguity and reasoning under uncertainty for multi-hypothesis dynamics in critical situations

Job Description

Missions

## Scientific context

Reasoning in critical situations often requires making decisions based on partial, ambiguous, or partially contradictory information, where several hypotheses may remain simultaneously plausible. In such contexts, prematurely stabilizing on a single interpretation may lead to fixation errors, particularly when discriminating information is absent, delayed, or perceived as unreliable.

While many simulation systems rely on scripted scenarios, few approaches address the formal generation of situations in which several diagnoses or interpretations must remain simultaneously plausible throughout the interaction. This raises several theoretical questions: how to explicitly represent partial informational states, how to maintain a controlled level of ambiguity, and how to organize the progressive introduction of elements that may redirect reasoning.

Uncertainty reasoning frameworks, in particular belief functions and related approaches, provide a relevant formal basis to address these questions, as they allow the explicit representation of ignorance, imprecision, and the coexistence of multiple compatible hypotheses.

---

# Scientific objective of the postdoctoral research

The objective of the postdoctoral research is to develop formal models enabling the automatic generation of ambiguous situations in immersive environments, based on mechanisms for reasoning under uncertainty.

A central challenge will be to formalize ambiguity not as a simple lack of information, but as the controlled maintenance of multiple compatible interpretations of the same situation. The goal is to produce situations that remain sufficiently open to preserve several reasoning trajectories, while maintaining a formal coherence compatible with the evolution of the scenario.

A second challenge will consist in maintaining this ambiguity within a credibility and plausibility range compatible with the progressive accumulation of information throughout the scenario.

The work will in particular investigate:

* how to maintain several compatible hypotheses while controlling the progressive introduction of discriminating information;
* how to control different degrees of ambiguity, incompleteness, or informational contradiction;
* how to generate informational configurations leading to differentiated inferential trajectories.

The postdoctoral research may also explore reasoning mechanisms taking into account different levels of representation of a given situation, depending on what is accessible, inferred, or ignored by the operator.

A particular perspective will consist in articulating these mechanisms with observables derived from the operator's activity, in order to dynamically adapt informational configurations to the hypotheses that may be privileged or neglected during the interaction.

This may include adjusting the dynamics of ambiguity according to the observed reasoning trajectories, by modulating the duration, intensity, or nature of the ambiguity maintained.

---

# Proposed scientific approach

The proposed approach relies in particular on the ability to:

* produce inconsistencies between different representations or information sources related to the same situation;
* temporarily hide relevant information in order to maintain a state of incompleteness;
* introduce variations in the reliability or credibility of information sources in order to generate states of doubt or informational blur.

These mechanisms should make it possible to explicitly control the emergence, maintenance, and progressive resolution of ambiguity throughout the scenario.

---

# Application context

The proposed research will contribute to the design of XR scenarios aimed at studying fixation errors in healthcare and aviation, and to the exploration of assistance mechanisms that may help detect or prevent such errors.

---

Activity

# Expected scientific contributions

Possible research directions include:

* extending belief-function-based models for the dynamic generation of ambiguous situations;
* computing informational configurations compatible with multiple competing diagnoses;
* formalizing mechanisms for selecting, masking, or degrading discriminating information;
* studying update mechanisms that control the evolution of credibility and plausibility of a scenario as successive information is introduced;
* exploring formalisms articulating logical coherence, uncertainty, and the concurrent evolution of multiple interpretive trajectories.

---

# Main tasks

* develop formal models for the automatic generation of ambiguous situations;
* contribute to their integration within interactive XR scenarios;
* participate in the scientific dissemination of the results.

---
## Documents required to apply

Send to domitile.lourdeaux@hds.utc.fr

* Curriculum vitae
* Motivation letter
* At least two references and/or recommendation letters
* A statement of research experience and interests

Any question about the position can be asked to the same addresses.

Your Profil

Skills

The position is intended for a candidate holding a doctorate in:

* artificial intelligence;
* computer science;
* applied mathematics;
* or a related discipline with a strong formal-modelling component.

Solid experience in several of the following areas will be particularly valued:

* reasoning under uncertainty;
* transferable belief model (TBM), belief functions, or evidence theory;
* knowledge representation;
* consistency-based diagnosis, abductive or multiple-hypothesis reasoning;
* logical or formal modelling;
* decision support;
* formal scenario generation.

Publications in at least one of these areas will be regarded as a particularly valuable asset.

An interest in questions situated at the interface between symbolic modelling, interpretive dynamics, and the understanding of human decision-making processes will be particularly appreciated.

An ability to combine a strong theoretical contribution with validation in an interdisciplinary framework applied to critical situations will constitute an important asset.

Your Work Environment

The Heudiasyc laboratory was created in 1981. Since its creation, it has been closely associated with CNRS and affiliated with the INS2I institute (Information Sciences) of CNRS.

Research conducted at Heudiasyc falls within the field of information and digital technologies (computer science, control, robotics, artificial intelligence). It aims at developing methods for representation, analysis, and control of systems subject to scientific, technological, economic, or social constraints.

Research activities are organized around three teams:

- CID: Knowledge, Uncertainty, Data
- SCOP: Reliability, Communication, Optimization
- SyRI: Robotic Systems in Interaction

The recruited researcher will join the CID research team, which is specialized in artificial intelligence.

The team's research focuses in particular on statistical learning, uncertainty management, and knowledge engineering, with applications in knowledge management, recommendation systems, and virtual environment scenario design.

The work will be carried out under the scientific responsibility of Domitile Lourdeaux and Sébastien Destercke.

The recruited researcher will also work in close collaboration with researchers from the Laboratoire Interdisciplinaire des Sciences du Numérique (LISN) and the Laboratoire Méthodes Formelles (LMF) at Université Paris-Saclay, in particular with Nicolas Sabouret, coordinator of the IDEFIX project, and Frédéric Boulanger.

Constraints and risks

None

Compensation and benefits

Compensation

€3,043 gross per month

Annual leave and RTT

44 jours

Remote Working practice and compensation

Pratique et indemnisation du TT

Transport

Prise en charge à 75% du coût et forfait mobilité durable jusqu’à 300€

About the offer

About the CNRS

The CNRS is a major player in fundamental research on a global scale. The CNRS is the only French organization active in all scientific fields. Its unique position as a multi-specialist allows it to bring together different disciplines to address the most important challenges of the contemporary world, in connection with the actors of change.

CNRS

The research professions